GEOSPATIAL ANALYSIS OF LATE PALEOINDAN HI-LO POINTS IN ONTARIO AND NEW YORK: TESTING EXPECTATIONS OF THE SETTLING IN HYPOTHESIS
A Thesis Submitted to the Committee on Graduate Studies in Partial Fulfillment of the Requirements for the Degree of Master of Arts in the Faculty of Arts and Science
TRENT UNIVERSITY Peterborough, Ontario, Canada (c) Copyright by Liam Richard Browne 2015 Anthropology M.A. Graduate Program May 2016
ABSTRACT
Geospatial Analysis of Late Paleoindian Hi-Lo Points in Ontario and New York: Testing Expectations of the Settling In Hypothesis
Liam Richard Browne
This thesis analyzes variability in a sample (n=302) of late Paleoindian Hi-Lo points from Ontario and New York. Biface variability is recorded using landmark geometric morphometrics. Raw material data is used to assess Hi-Lo toolstone usage patterns and the impact of raw material constraints on manufacture. Statistical analyses are used to assess patterning of variability in space.
Spatial results are interpreted using cultural transmission theory in terms of their implications for the geographic scale of social learning among Hi-Lo knappers. Results of the spatial analyses are related to theory about hunter-gatherer social networks in order to understand the effects of hypothesized settling in processes on late Paleoindian knappers.
Results indicate random spatial patterning of Hi-Lo variability. The absence of spatial autocorrelation for Hi-Lo size indicates that settling in processes were not sufficiently pronounced during the late Paleoindian period to manifest as inter-regional variability within the Hi-Lo type.
Keywords: Ontario Archaeology, New York Archaeology, Hi-Lo, Late Paleoindian, Settling In,
Biface Variability, Geometric Morphometrics, Cultural Transmission Theory, Raw Material
Usage, Mobility, Early Holocene
ii
Acknowledgments
I first like to thank my supervisor Dr. James Conolly for the project idea, as well as his guidance and support during this research. I would also like to acknowledge the important editorial contributions of my committee members, Dr. Laure Dubreuil, William Fox and Dr. Chris Ellis.
In particular I would like to acknowledge the efforts of Dr. Brian Deller, William Fox and Steven
Timmermans to support my research. Each went above and beyond the call of duty to help me seek out lesser known Hi-Lo points throughout Ontario. I am also grateful to Robert von Bitter and the Ministry of Tourism, Culture and Sport for providing a site records form detailing known
Hi-Lo finds in Ontario. This served as my initial guide to the world of Ontario Hi-Lo points.
Without the help of these individuals my search for Hi-Lo may never have gotten off the ground.
I would also like to thank Dan Long and Fred Moerschfelder for taking the time to share their knowledge of Ontario cherts with me. Michael Martyniuk for assisting with point illustrations and Ramsay MacFie for accompanying me on a number of late night Hi-Lo road trips.
This research made use of archaeological data that were obtained from collections held by a variety of institutions, consulting archaeology firms and private individuals. I am greatly indebted to those who enabled access to collections or took the time to pass along relevant archaeological reports as I tracked down old references. I am grateful to following entities and individuals who facilitated access to collections: Archaeological Research Associates (Andrea
Carswell and Paul Racher), Archaeological Services Inc. (Caitlin Coleman), Clarington Museums
(Charles Taws), D.R. Poulton and Associates (Dana Poulton), Fisher Archaeological Services
(Jacqueline Fisher), Golder Associates (Jamie Davidson), the Ministry of Transportation of
Ontario (Dr. Robert Pearce), the Museum of Ontario Archaeology (Joan Kannigan), Northeastern
iii
Archaeological Associates (Dr. Lawrence Jackson), the Port Colbourne Museum (Stephanie
Baswick), the Port Perry Museum, the Regional Municipality of Waterloo, the Royal Ontario
Museum (April Hawkins), Sustainable Archaeology – London (Dr. Rhonda Bathurst and Kyra
Westby), Timmins-Martelle Heritage Consultants (Dr. Holly Martelle, Dr. Peter Timmins and
Nicole Brandon), the Toronto and Region Conservation Authority (Margie Kennedy), Western
University (Dr. Chris Ellis) and York North Archaeological Services (Gordon and Patricia Dibb).
I would also like to acknowledge the contributions of all the private individuals who enabled the use of private artifact collections in this study, whether by directly providing access to collections or by making arrangements to facilitate access. In many cases, these people were gracious enough to invite me into their home and share with me their ideas, knowledge and passion for archaeology. The following individuals are sincerely thanked for their contributions to this study:
Richard Baskey, Dr. Brian Deller, Douglas Dennis, Ilse Kraemer, Laura Lee, Dr. Roger Lorhman,
Dan Long, Fred Moerschfelder, Scott Oliver, Steve Timmermans and Robert Titmus.
Monetary support for this research was provided by the Trent University Archaeological
Research Centre (TUARC) and a Joseph Armand Bombardier Canada Graduate Scholarship -
Master’s provided by the Social Sciences and Humanities Research Council of Canada.
Finally I would like to thank my parents, Richard and Nancy for their support.
iv
Table of Contents
Abstract ii
Acknowledgements iii
Table of Contents v
List of Figures x
List of Tables xii
Chapter 1: Introduction 1
1 Introduction and Research Goals 1
1.2 Hi-Lo Polythetic Typological Definition 2
1.3 The Settling in Hypothesis and the Ontario Archaeological Record 3
1.4 Cultural Transmission Theory and the Settling in Hypothesis 4
1.5 Thesis Organization 5
Chapter 2: The Paleoindian Occupation of Southern Ontario 7
2 Introduction 7
2.1 Development of Paleoindian Research in Ontario 8
2.2 The Fluted Point Tradition and the Dating of Ontario’s 9 Paleoindian Biface Sequence
2.2.A Early Paleoindian 10
2.2.B Late Paleoindian 13
2.3 The Ontario Paleoindian Biface Sequence 13
2.4 The Hi-Lo Phase 15
2.4.A Type Description 16
2.4.B Hi-Lo Subtypes 17
2.4.C Hi-Lo Raw Material Usage Patterns in Ontario 19
2.4.D Distribution and Excavated Sites 23
2.5 Great Lakes Paleoindian Technology 24
v
2.6 Raw Materials Common in Ontario Paleoindian Assemblages 25
2.7 Early Paleoindian Raw Material Usage Patterns in Ontario 27
2.7.A Gainey Phase 27
2.7.B Parkhill Phase 28
2.7.C Crowfield Phase 29
2.8 Late Paleoindian Raw Material Usage Patterns in Ontario (Excluding Hi-Lo) 30
2.8.A Holcombe Phase 30
2.8.B Madina Phase 30
2.9 Paleoindian Mobility 31
2.9.A The High Technology Forager Model 32
2.9.B Northeastern Paleoindian Mobility and Caribou Predation 34
2.9.C Early Paleoindian Raw Material Procurement and Mobility 34
2.9.D Ontario Early Paleoindian Mobility and Caribou Predation 39
2.10 Implications of the Settling In Hypothesis for Late Paleoindian 40 Mobility and Raw Material Procurement
2.10.A Late Paleoindian Mobility and Caribou Predation 41
2.10.B Late Paleoindian Mobility and Raw Material Procurement 42
Chapter 3: Cultural Transmission Theory and Cultural Change 44
3 Explaining Biface Variability 44
3.1 Risk 45
3.2 Prey Availability 47
3.3 Mobility and Raw Material Management 49
3.4 Cultural Transmission Theory 51
3.4.A Modes of Transmission 51
3.4.B Guided Variation and Common Descent 52
3.4.C Reconstructing Heritable Continuity through 52 Archaeological Phylogenetic
3.5 Drift and Cultural Change 53
3.5.A Copying Error as a Source of Drift 54
vi
3.5.B Innovation Causing Drift 54
3.5.C Drift and Effective Population Size 55
3.6 Hypotheses 57
3.7 Selection (Risk/Function) 57
3.7.A Innovation as a Response to Risk 58
3.8 Drift: Accumulated Copying Error and Innovation in an 60 Unbounded Learning System
3.8.A Inherent Synchronous Design Variation 61
3.8.B Geographic Regionalization of Variation in an Internally 62 Bounded Learning System
3.9 Design Variation Driven by Raw Material Constraints 64
Chapter 4: Methods and Sample Composition 65
4 Quantitative Assessment of Shape and Size 65
4.1 Multivariate Morphometrics 66
4.2 Geometric Morphometrics 67
4.2.A Landmark Classification 68
4.3 Collection and Analysis of Metric Data 69
4.3.A 2015 Hi-Lo Sample and the Expanded 2015 Hi-Lo Sample 70
4.3.B Implications of Non-Projectile Functional Modifications of 76 Form for Sample Selection
4.4 2015 Hi-Lo Sample Provenience 77
4.5 Maximum Thickness 78
4.6 Weight 78
4.7 Raw Material Identification 78
4.8 Landmark Placement for Geometric Morphometric Analysis 79 of Hi-Lo Bifaces
4.9 Metric Variables 81
4.10 Character State Analysis 83
vii
Chapter 5: Results and Analysis 88
5 Results and Analysis 88
5.1 Type Description 88
5.2 Ontario and Western New York Hi-Lo Distribution 89
5.2.A Possible Occurrences of Hi-Lo in Eastern Ontario 90
5.2.B Western New York Hi-Lo Distribution 91
5.3 Raw Material Identification and Distribution 92
5.4 Regional Raw Material Analysis 97
5.5 Metric Variable Analysis 102
5.5.A Variable Correlation 104
5.6 Effects of Raw Material on Hi-Lo Design 105
5.6.A Coefficients of Variation for Variables by Raw Material 106
5.6.B Mann-Whitney U Test Analysis 112
5.7 External Coefficient of Variation Comparison 116
5.8 Principal Components Analysis 118
5.9 Whole Point PCA 120
5.9.A PC1 Whole Point 122
5.9.B PC2 Whole Point 124
5.9.C PC3 Whole Point 125
5.10 Hi-Lo Base PCA 126
5.10.A PC1 Base 128
5.10.B PC2 Base 129
5.10.C PC3 Base 130
5.11 Spatial Patterning in Hi-Lo Variability 132
5.12 Moran’s Index 132
5.12.A Results 133
5.13 Mantel Correlogram 134
5.13.A Results 135
5.14 Regional PCA Analysis 139
viii
5.14.A Results 139
5.15 Chapter Summary 142
Chapter 6: Conclusions 143
6.1 Future Directions for Research 146
6.2 Research Evaluation 147
6.3 Research Value 148
6.4 Final Remarks 149
References Cited 150
Appendices (on accompanying CD)
A Site List and Compiled Hi-Lo Variabile Data
ix
List of Figures
Figure Description Page 1 Select Paleoindian sites and chert sources mentioned in the text 1
2 Ontario and eastern Great Lakes Paleoindian biface sequence 13
3 Hi-Lo subtypes proposed by Ellis 2004a 18
4 Comparison of maximum length and width between 67 oval and teardrop shape
5 Hi-Lo collections accessed 70
6 Study areas and Hi-Lo raw material identifications in Roberts 1985 71
7 Summary of Hi-Lo documented in Tinkler and Pengelly 2004 showing 74 site locations included in 2015 Hi-Lo Sample
8 Hi-Lo displaying tip impact. Note the false flute resulting from tip impact 76
9 Hi-Lo displaying extensive beveling on a single lateral edge possibly 76 due to functional modification for non-projectile use as a side scraper
10 Hi-Lo displaying extensive beveling on a single lateral edge possibly due 76 functional modification for non-projectile use as a side scraper
11 Hi-Lo with extensive longitudal resharpening. Note the 76 blunted tip suggesting non-projectile use as an end scraper
12 Hi-Lo Point of Haldimand chert, Stelco I Site (AfHa-200) 79
13 Hi-Lo Point of Kettle Point chert, Leslie Site (AhGw-32) 79
14 Hi-Lo Point of Onondaga chert, Cox II Site (AfGt-63) 79
15 Hi-Lo Point of Bayport chert, Welke-Tonkonoh Site (AfHj-5) 79
16 Selection of metric variables included in analysis 81
17 Contracting ear angle measurement (P) 86
18 Character states adapted from O’Brien et al. 2014 87
19 Ontario and western New York Hi-Lo distribution 89 (Expanded 2015 Hi-Lo sample distribution)
20 2015 Hi-Lo sample distribution 90
21 Grenadier Island biface 90
22 Haldimand Hi-Lo distribution 93
x
23 Onondaga Hi-Lo distribution 94
24 Kettle Point Hi-Lo distribution 94
25 Seaton Lands Hi-Lo - Block H (Findspot 2) 95
26 Bayport chert Hi-Lo - Elgin County, Ontario 97
27 Expanded 2015 Hi-Lo sample raw material by geographic region 100
28 2015 Hi-Lo Sample raw material by geographic region 101
29 Coefficient of variation for Hi-Lo variables by raw material group 107
30 Coefficient of variation for equivalent measurement by Hi-Lo study 117
31 Whole Point PCA - principal components 1 versus 2 121 and principal components 1 versus 3
32 Hi-Lo Whole Point PCA - principal component 1 123
33 Hi-Lo Whole Point PCA - principal component 2 125
34 Hi-Lo Whole Point PCA - principal component 3 126
35 Hi-Lo Base PCA - principal component 1 versus 2 127 and principal component 1 versus 3
36 Hi-Lo Base PCA - principal component 1 129
37 Hi-Lo Base PCA - principal component 2 130
38 Hi-Lo Base PCA - principal component 3 131
39 Mantel correlogram expressing types of spatial variability 135
40 Hi-Lo location heat map 136
41 Mantel correlograms for PC1-3 of Whole Point PCA 137
42 Mantel correlograms for PC1-3 of Hi-Lo Base PCA 138
43 Whole Point PCA - principal components 1 versus 2 140 and principal components 1 versus 3 sorted by region
44 Hi-Lo Base PCA - principal components 1 versus 2 141 and principal components 1 versus 3 sorted by region
xi
List of Tables
Table Description Page 1 Attributes used in polythetic definition of Hi-Lo 2
2 Hi-Lo sites excavated in Ontario 23
3 Roberts (1985) Hi-Lo sites plotted in Fig 19. Ontario and New York 73 Hi-Lo distribution (expanded 2015 Hi-Lo sample)
4 Tinkler and Pengelly (2004) Hi-Lo included in 2015 Hi-Lo sample 74
5 Hi-Lo Sites plotted in Fig 19. Ontario and New York 75 Hi-Lo distribution (Included in Expanded 2015 Hi-Lo Sample)
6 Landmark positions and descriptions 80
7 Metric variables defined by horizontal differences 82
8 Metric variables defined by vertical differences 82
9 Character state analysis adapted from O’Brien at al. 2014 84
10 Angle measurement - straight and expanding tangs (90 degrees or less) 85
11 Angle measurement - contracting tangs (< 90 degrees) 85
12 Raw material affiliation (expanded 2015 Hi-Lo sample) 92
13 Raw material affiliation (2015 Hi-Lo sample) 92
14 Summary descriptive statistics for total Hi-Lo sample (n=302) 102
15 Summary descriptive statistics for Haldimand Hi-Lo sample (n=133) 102
16 Summary descriptive statistics for Kettle Point Hi-Lo sample (n=49) 103
17 Summary descriptive statistics for Onondaga Hi-Lo sample (n=62) 104
18 Correlation coefficients of a magnitude >0.80 105
19 Coefficients of variation for 2015 Hi-Lo sample (n=302) 107
20 Variable coefficients of variation showing significant differences 108 between raw material groups
21 Mann-Whitney U test results for Hi-Lo variables by raw material 113
22 Coefficient of variation comparison for Hi-Lo 116
23 Rotated eigenvector values for principal components 1, 2 and 3 120 for Whole Point PCA
xii
24 Rotated eigenvector values for principal components 1, 2 and 3 126 for Hi-Lo Base PCA
25 Moran’s I results for Whole Point PCA 134
26 Moran’s I results for Hi-Lo Base PCA 134
xiii
1
Chapter 1
Introduction
1.0 Introduction and Research Goals
The study of stone tools and their manufacture is an enduring avenue of research in anthropological archaeology. Archaeologists employ typologies to express the concurrence of variants within classes of lithic bifaces in operational terms (O’Brien and Lyman 2002: 38).
Typologies are meant to express a range of variability in artifact form which is considered to form a heuristic unit with explanatory significance (O’Brien and Lyman 2002). Most types as employed in Paleoindian archaeology today can be considered extensional units; this is because their definitions stem from initial type sites where the particular form was ‘recognized’ as a type.
Extensional type definition comprise the attributes shared by members of the unit; where those attributes defining the unit were initially formulated based on observed attributes of members of the unit (O’Brien and Lyman 2002). Since most Paleoindian type definitions were derived from eponymous type site collections, their value rests on the assumption that the types thus defined have significance as cultural-temporal markers. If this assumption is correct, variability expressed by members within a type should represent differential expressions of cultural information over a period of time; some of this information can be recognized by archaeologists today.
This research uses landmark geometric morphometrics to analyze a sample (n=302) of Hi-Lo type lithic projectile points from Ontario and New York. The type was initially extensionally defined in 1963 by James Fitting based on an assemblage recovered from the Hi-Lo Gun Club in
Michigan. The large assemblage of Hi-Lo points reported by Ellis and Deller (1982; 1990; 2014) from the Welke-Tonkonoh site (AfHj-5) has been very important for the definition and study of this type in Ontario.
2
1.2 Hi-Lo Polythetic Type Definition
Ellis and Deller (2014, 33) have suggested that a polythetic (see Clarke 1968) type characterization is necessary to in order to properly distinguish Hi-Lo from similar, and perhaps ancestral, lower Great Lakes point forms such as Holcombe (see Ellis 2004a). A polythetic approach to Hi-Lo typological definition advocates the use of the variation in attribute states to distinguish Hi-Lo from other similar forms. This approach is meant to expand the original normative definition of Fitting (1963) which is largely composed of invariant attributes. When necessary, typological identifications were made with reference to the work of Ellis and Deller
(1982, 2014). The attributes included in the polythetic definition of Ellis and Deller (2014: 33) are presented, and ordered according to their referent, in Table 1.
Table 1. Attributes used in polythetic defintion of Hi-Lo
Point Morphology Blade Element Haft Element Marked plano-convex cross- Thick blades Distinct lateral basal thinning section from one edge Pronounced longitudinal Very blunt tips Large thick ears/basal corners curvature Large surface areas covered by Edge beveling Incurvate to slightly notched original flake blank remnants basal side edges Shoulders or stems
Hi-Lo points date to approximately 10,400-10,000 BP during a time of major cultural and environmental change (Ellis et al. 2009). While many aspects of the late Paleoindian period are poorly understood late Paleoindian archaeology is an active area of research in Ontario and the wider Great Lakes region.
In this thesis I propose to answer the following questions with regards to Hi-Lo technology in
Ontario and New York:
1. What is the geographic distribution of Hi-Lo points in Ontario and New York?
3
2. What raw materials were preferred for Hi-Lo point manufacture in the study area?
3. How is the use of different raw materials, each with variable workability characteristics,
reflected in Hi-Lo design and manufacutre?
4. Are established ideas of Hi-Lo raw material usage in Ontario (e.g. Ellis 2004a; Ellis and
Deller 1982, 2014) supported by the current Hi-Lo record?
5. Which aspects of Hi-Lo size (e.g. basal concavity depth or maximum basal width) best
characterize the range of variability within the type?
6. What sort of geographical patterning can be observed for Hi-Lo point variability?
7. If Hi-Lo point variability is seen as the material product of a learning system shared
between hunter-gather groups within the study area, how does spatial patterning in
variability relate to processes of settling in which are thought to have been ongoing at the
time when Hi-Lo points were in use.
1.3 The Settling in Hypothesis and the Ontario Archaeological Record
Change in material cultural following the initial Paleoindian colonization of Ontario (circa 11,000
BP) is often related to long term processes of settling in to the landscape. The settling in model is based upon postulated decreases in mobility for hunter-gatherer groups subsequent to initial exploration and colonization (Ellis 2011: 387). As time passed, Paleoindian colonists would have become more familiar with their adopted landscape. The process of familiarization is thought to have encouraged decreasing in residential mobility and the broadening of subsistence strategies to include the exploitation of food or toolstone resources which previously were either unknown or ignored (Ellis et al. 1990: 66). Unfortunately, the lack of detailed floral or faunal remains associated with Hi-Lo occupations makes it difficult to directly analyze potential changes in subsistence at this time. However, it is possible that settling in processes and the shift from cooler
4 late Pleistocene conditions to warmer and wetter Holocene environments (10,000 BP – present) may have encouraged population growth during the time when Hi-Lo points were in use (Karrow and Warner 1990; Ellis et al. 1990: 67).
Across eastern North America, the early Paleoindian fluted point tradition is supplanted in the archaeological record by stylistically diverse late Paleoindian unfluted point types with more limited geographic distributions (White 2013). In the Central Mississippi Valley, the regional stylistic variability during the late Paleoindian period has been interpreted as a result of the development social boundaries, which act as inhibitors to the transmission of cultural information, through processes of settling in (Koldehoff and Walthall 2004, 2009; Koldehoff and Loebel 2009;
Walthall and Koldehoff 1998).
In the Northeast, significant regional variation in manufacture has been demonstrated to exist for
Laurentian Archaic point forms of the middle Archaic (8,000-4,500 BP) (Conolly 2015). Further diversification of point forms is seen in late Archaic (4,500-2,800 BP). This period is noted for a higher degree of variability between roughly contemporaneous sites and the use of a wide range of projectile point forms (Ellis et al. 1990: 39; Ellis 2011; Wright 1995: 217). While a great expanse of time separates these periods from Hi-Lo, they may represent variability at different stages of the settling in process.
1.4 Cultural Transmission Theory and the Settling in Hypothesis
Understanding the impetus for cultural change has long been a vital part of anthropological archaeology (Collard et al. 2005, 2011; Eerkens et al. 2014; Shennan 2001, 2002; O’Brien et al.
2014). Cultural transmission theory provides a framework for the interpretation of variability in the Hi-Lo type and explains variability in terms of the geographic scale of social learning for knappers. Decreased residential mobility and assertions of territoriality associated with settling in processes are predicted by cultural transmission theory to encourage the proliferation of local
5 design variants within the larger Hi-Lo type (Ellis 2011: 394; Smallwood 2013: 692). If Hi-Lo points in Ontario and New York were produced at a time of significant social network fragmentation due to settling in, regional sub-populations could be expected to develop different design variants as their subsistence strategy became more directly conditioned by local environmental conditions. Decreased mobility and territoriality would reduce the technologically homogenizing effects of inter-group social learning between groups of geographically dispersed
Hi-Lo knappers (White 2013b; Henrich 2004; Eerkens et al. 2014).
This research describes the scale of Hi-Lo social learning as a means to test for the existence of predicted archaeological manifestations of settling in processes. Cultural transmission theory predicts that the existence of settling in processes at this time might be recognized through non- random spatial patterning in Hi-Lo variability. This thesis tests the primary hypothesis that Hi-Lo knappers in Ontario and New York inhabited a social network characterized by regional sub- populations whose degree of interaction was limited.
1.5 Thesis Organization
This thesis is divided into six chapters:
Chapter 2 The Paleoindian Occupation of Southern Ontario provides a literature review of relevant Paleoindian research focusing on southern Ontario and the eastern Great Lakes. This chapter reviews the development and support for the relative sequence of biface forms used by archaeologists to order technological developments in Ontario’s Paleoindian period. Lithic procurement patterns and proposed hunting strategies are discussed as they relate to estimates of
Paleoindian mobility. Evidence relating to both early and late Paleoindian cultural phases is reviewed to provide context for this work’s analysis of the Hi-Lo bifaces. In addition to type description, this chapter provides a review of the evidence for Hi-Lo lithic procurement patterns and mobility.
6
Chapter 3 Cultural Transmission Theory and Cultural Change reviews developments in evolutionary archaeology and cultural transmission theory pertinent to this study. This chapter emphasizes theoretical support for mechanisms of cultural change. The chapter concludes by presenting hypotheses informed by cultural transmission theory in the form of best-fit scenarios to describe the scale of social learning during the time Hi-Lo points were produced.
Chapter 4 Methods and Sample Composition introduces the reader to the analyses of size and shape and describes the technique of landmark geometric morphometrics which is used in this study for the assessment of Hi-Lo point size. Variables used in the analysis are presented and described. The Hi-Lo sample used in this study is further described; highlighting point provenience information and raw material classifications. This chapter importantly introduces a number of points attested in the literature for which provenience and raw material affiliation is known but whose size was not able assessed as part of this research.
Chapter 5 Results and Analysis presents results of the analysis detailed in Chapter 4 and statistical analyses used to assess the scale of social learning during the Hi-Lo period. This chapter presents point metrics describing aspects of Hi-Lo size and configuration and evaluates the influence of raw material constraints on Hi-Lo design. Raw material preferences of knappers creating Hi-Lo points are assessed and results are contrasted with previous regional studies of Hi-Lo points.
Chapter 6 Conclusions provides final conclusions drawn from this analysis. Hypotheses and research goals are revisited and evaluated. This section includes discussion of problems with the sample or methods used in this analysis and provides some directions for future Hi-Lo research.
7
Chapter 2
The Paleoindian Occupation of Southern Ontario
2 Introduction
The purpose of this chapter is to provide an overview of Paleoindian archaeology in southern
Ontario. A main focus will be the development and support for the chronological sequence of bifaces used to order Paleoindian technological variants in Ontario (Deller and Ellis 1988; Ellis and Deller 1990). This chapter will discuss characteristics of Paleoindian technology, subsistence and mobility patterns and highlight change over time. Locations of relevant Paleoindian sites and chert sources are shown by Figure 1.
Figure 1. Select Paleoindian sites and chert sources mentioned in the text
8
2.1 Development of Paleoindian Research in Ontario
Paleoindian point forms in Ontario were first recognized by William Patterson, a geology student at the University of Western Ontario (Jackson et al. 1987). Although, fluted points had been previously depicted in the Ontario literature their significance was not recognized then (Boyle
1906: 11). Despite the assurances of Jesse Figgins, discoverer of the Folsom site, the academic establishment of the day ignored Patterson’s finds due to general disinterest an unwillingness to challenge the accepted literalist interpretation of the Judeo-Christian creation narrative. A reading of the correspondence of Patterson and Figgins suggest that the Canadian academic community was “not prepared to accept evidence of Late Pleistocene man within its borders” (Jackson et al.
1987: 5). In 1951, Kenneth Kidd (1951) published a small description of eleven fluted points from
Ontario. This is commonly regarded as the beginning of Paleoindian studies in Ontario rather than
Patterson’s earlier discoveries (Jackson et al. 1987) Later researchers in the 1960’s and 70’s drew attention to the wealth of Paleoindian materials in southern Ontario (Deller 1976, 1979; Deller and Ellis 1992a; Garrad 1971; Jackson 1983; Roosa 1977; Storck 1973; 1979).
William Roosa (1965) was the first to suggest the possibility of a temporal sequence of fluted points in the Great Lakes. He emphasized differences in fluting techniques between Paleoindian point forms as a means to delineate types. Roosa was among the first to consistently emphasize the importance of understanding temporal and developmental relationships between Paleoindian point forms. He saw this as an essential part of developing typologies with explanatory significance. His emphasis on Great Lakes projectile form sequences has led to the development of one of the most sophisticated point sequences in eastern North America (Ellis 1996). Roosa’s work at both the Barnes site in Michigan (Wright and Roosa 1966) and the Parkhill site in Ontario
(Roosa 1977) helped to demonstrate the spatial distribution of similar point forms in the Great
Lakes (Jackson 1998).
9
A major development in Paleoindian research came with Charles Garrad’s (1971) survey of fluted points. Garrad’s work introduced a great number of previously unknown fluted point find-spots and some sites to researchers. The depiction of a number of these points in Garrad’s publication shed further light upon the variation found in southern Ontario fluted point forms. His work also revealed a concentration of fluted points in southwestern Ontario. The discovery of Clovis-like
Gainey points at the type site in Michigan in the early 1980’s brought Great Lakes and Ontario fluted point variation into clearer focus (Simons et al. 1984). Gainey points and distinctive fluted points from the Barnes site (Roosa 1977) and the Crowfield site (Deller and Ellis 1984, 2011;
Ellis 1984c) are considered today to be the principal manifestations of the eastern fluted point tradition in southern Ontario.
2.2 The Fluted Point Tradition and the Dating of Ontario’s Paleoindian Biface Sequence
Our understanding of the chronology of the Paleoindian period in Ontario is heavily reliant on the sequencing of biface forms. A well-developed relative sequence of Paleoindian point types has been developed for southern Ontario and archaeologically similar areas of the eastern Great Lakes
(Ellis and Deller 1990, 1997; Deller and Ellis 1992b; Lothrop and Bradley 2012). The sequence is commonly divided into early and late Paleoindian manifestations.
(1) The early Paleoindian period is defined by lanceolate fluted points. Fluting refers to the removal of distinctive long channel flakes on one or, more commonly, both faces of a projectile point (Ellis and Deller 1990). Early Paleoindian points are also noted for their lanceolate shape, concave bases and grinding of the lower and basal portions.
(2) The late Paleoindian period is also defined on basis of changes in point morphology (Ellis and Deller 1990; Meltzer 2009). Late Paleoindian points exhibit many of the same characteristics as early Paleoindian forms; a general lanceolate outline, basal grinding and an absence of
10 notching. These points can be broadly distinguished from early Paleoindian point forms by their lack of fluting (Ellis and Deller 1990).
To order these biface types in a chronological sequence researchers have looked to securely dated and seriated Paleoindian point types from elsewhere in North America. In absence of a robust radiocarbon record for eastern Great Lakes Paleoindian sites (Boulanger and Lyman 2014) researchers have constructed a relative chronological framework for the period based on inferred developmental relationships between diagnostic biface forms.
Absolute chronometric dating for Ontario’s Paleoindian period can be inferred on the basis of perceived cultural and temporal relations to securely dated sites elsewhere in North America (e.g.
Ellis and Deller 1990; Dibb 2004). Within Ontario, additional temporal definition for the period is provided by associations of diagnostic artifact types with securely dated geological events
(Karrow et al. 1975; Karrow and Warner 1990; Deller, 1976, 1979; Jackson et al. 2000; Storck
1984a, 1984b; Morgan et al. 2000).
2.2.A Early Paleoindian
The eastern North American fluted point tradition includes a wide range of regional variants (Ellis and Deller 1990; O’Brien et al. 2014; Thulman 2012; White 2006; Macdonald 1968; Simons et al.
1984; Wright and Roosa 1966; Grimes 1979). The Ontario biface sequence is anchored by the assumption that point forms most similar to Clovis points are oldest and subsequent changes in design represent deviations from an initial Clovis primogeniture (Deller and Ellis 1992a; White
2006).
Clovis sites are the earliest securely dated fluted point sites in North America. Excavated Clovis sites have been dated to between c.a 11,500- 11,000 BP in western North America (Fiedel 1999;
Goebel et al. 2008; Haynes 1980; Haynes et al. 1984, 2007; Meltzer 2009; Ferring 2001). The date range of Clovis has been challenged in recent years by Waters and Stafford (2007) whose
11 reassessment of the radiocarbon record places Clovis from 11.200-11.100 to 10.900-10.800 BP.
Waters and Stafford’s date range has been met with skepticism by some and is not unanimously accepted (Haynes et al. 2007).
Clovis points are bifacially flaked, lanceolate points with parallel to slightly convex sides and concave bases. A defining feature of the Clovis point is the presence of a fluting scar resulting from the removal of a long channel flake. Clovis flutes often extend about one third of the way to the point’s tip. (Buchanan and Collard 2010; Howard 1990; O’Brien et al. 2014; Wormington
1957).
Clovis technology is assumed to have originated among a population in the interior of North
America around 12,000 BP (Buchanan and Collard 2007; Faught 2008; Goebel et al. 2008).
Fluted point technology subsequently spread through colonization and/or an existing population.
Although the majority of fluted points finds are located in eastern North America; the spatial gradient of Clovis-age radiocarbon dates may suggest a west-east spread of fluted point technology (Hamilton and Buchanan 2007).
Southern Ontario was largely inaccessible for human occupation prior to 12,000-11,500 BP due to glacial ice and impoverished, inhospitable environments (Karrow and Warner 1990). Few sites in the Northeast have provided dates much older than 10,700 BP (Miller and Gingerich 2013).
Thermoluminescence dates from Clovis-like Gainey site in Michigan (Simons 1997) suggest an occupation age of c.a. 11,000 BP. Ellis and Deller (1988, 1990) have suggested the early
Paleoindian period in Ontario dates to c.a. 11,000 to 10,500 BP.
Within Ontario, no fluted points have been found in clear contexts below the strandlines of glacial
Lake Algonquin (Deller 1976, 1979; Storck 1982; Jackson 1983; Roberts 1985; Ellis and Deller
1986, Deller and Ellis 1992a, 1992b) which is thought to have drained around 10,400 BP (Karrow et al. 1975; Jackson et al. 2000). Since numerous early Paleoindian sites have been discovered on,
12 or associated with the landward side of the Algonquin strandlines, fluted point use is thought to coincide with high water levels in the Huron basin (Ellis and Deller 1986).
2.2.B Late Paleoindian
Across North America unfluted forms succeed fluted points and a similar relationship is assumed in Ontario. The later temporal placement of the late Paleoindian forms in the Ontario biface sequence is supported by geological evidence and presumed common temporal/developmental relationships with unfluted point forms elsewhere in North America. .
In western North America, particularly on the Great Plains and in the Southwest, Clovis points are succeed by fluted Folsom points (Fiedel 1999). Radiocarbon dates place the inception of Folsom at c.a. 10,900 BP in western North America (Fiedel 1999: Fig. 6) Following Folsom, point forms in western North America largely cease to be fluted and unfluted late Paleoindian points come to dominate assemblages (Meltzer 2009). In western North America and the Southeast, late
Paleoindian point forms are known to have been in use from 10,400 – 10,000 BP (Frison 1976;
Frison and Stanford 1982; Goodyear 1982; Holliday et al. 1983; Holliday 2000; Pitblado 2003:
Table 5.12).
In Ontario, late Paleoindian points have been found on the bed of glacial Lake Algonquin while early Paleoindian forms are found only on the landward sides of the strandlines of the former lake. Since Algonquin is thought to have drained around 10,400 BP with the opening of the
Kirkfield Outlet (Karrow 1975; Ellis and Deller 1986) 10,400 BP can be used as estimate for the maximum potential age of Hi-Lo and Madina points. Further comparison of Ontario Paleoindian point forms to well-dated late Paleoindian point types in other areas suggests they pre-date ca.
9,500 BP (Dibb 1985:236; Ellis and Deller 1986:55).
13
2.3 The Ontario Paleoindian Biface Sequence
The Paleoindian point sequence for southern Ontario has been defined by Ellis and Deller (1986;
1988; 1990; 1992a). This relative chronological sequence has found acceptance among many
Great Lakes researchers (e.g. O’Brien et al. 2014; Bradley and Lothrop 2012; Lothrop et al.
2011), however, some researchers (e.g. White 2012) propose developmental relationships between the point forms.
Figure 2. Ontario and eastern Great Lakes Paleoindian biface sequence
For fully fluted biface forms the relative sequential chronology is Gainey-Barnes-Crowfield
(Deller and Ellis 1988; Ellis and Deller 1990; Lothrop and Bradley 2012). These point types are used as basic index fossils for larger cultural units known as complexes or phases. Cultural phases linked to these point types are known as Gainey, Parkhill and Crowfield respectively (Deller and
Ellis 1988, 1992a; Roosa 1977; Shott 1986; Storck 1984b).
Deller and Ellis (1988) regard the bifaces types as “arbitrary segments in a continually changing system, each of which is sufficiently differentiated in time to be recognized as a separate type.”
14
Given the paucity of radiocarbon dates for eastern Great Lakes Paleoindian components such a conception is prudent. Any proposed absolute temporal ranges for each phase of the Paleoindian period should be regarded with caution until such a time when a robust radiocarbon dataset becomes available.
The idea that these point types can act as temporal markers for phases of cultural evolution is supported by multiple of lines of reasoning. Paleoindian point types are found through large and frequently overlapping areas of the eastern Great Lakes (e.g. Roosa 1977; Simons et al. 1984;
Lothrop and Bradley 2012; Ellis et al. 2011). However, despite the overlapping distributions of these points, most sites only include a single point type (Deller and Ellis 1988; Simons et al.
1984). This evidence suggests that each point form exists independently of the others.
Each point type displays substantial differences in form (e.g. Gainey: Simons et al. 1984; Ellis
1984a, Barnes: Ellis 1984b, Crowfield: Ellis 1984c) which have been interpreted as intentional design choices. Of the three fluted point types in Ontario, Gainey and Crowfield show the greatest dissimilarity to each other while each shows the greatest degree of similarity to Barnes. This is considered as evidence for an intermediate placement for Barnes the developmental sequence
(Deller and Ellis 1992b).
Design considerations imposed by raw material constraints have been considered and rejected as a causal factor driving fluted point variability in Ontario (Deller and Ellis 1992b). Gainey, Barnes and Crowfield points have all been observed to occur on the same types of raw materials (e.g.
Collingwood chert of the Fossil Hill formation). The ability of Paleoindian knappers to manufacture each of these point forms from Collingwood chert suggests that raw material constraints cannot account for the observed differences in fluted point design. Furthermore, Roosa
(1963, 1965) found that resharpening could not explain morphological differences between the early Paleoindian point types.
15
Late Paleoindian unfluted forms (Holcombe; Ellis 1986; Fitting et al. 1966) (Hi-Lo; Fitting 1963;
Ellis 1981) (Madina; Dibb 1985, 2004; Dibb and Ellis 1988: 20) also have significant differences in form. While, Hi-Lo and Holcombe were used in the same geographic area, most sites show evidence for only occupation by peoples using points of a single type with multicomponent sites being considered as palimpsests. Since there is no discrete (i.e. non-overlapping) spatial patterning displayed by sites yielding Holcombe and Hi-Lo points, the types are not thought to be contemporaneous. Instead, they are considered to be later design variants related to the developmental sequence of fluted points by virtue of the apparent continuity between Crowfield and Holcombe and subsequently Holcombe and Hi-Lo (Ellis and Deller 1990: 57).
White (2012: 198) sees the fluted bifaces from the Crowfield site as similar to many other points which are included within the Holcombe type. In this case, perceived continuity may be a product of similar typological definitions for Crowfield and Holcombe points.
Developmental continuity between Holcombe and Hi-Lo is not accepted by all researchers. White
(2006: 50, 2012: 343) sees the appearance of Hi-Lo in the Great Lakes as representative of a population expansion, possibly via the Illinois River valley, of peoples associated with the Dalton horizon. The expansion of these groups northeast was facilitated by an environmental shift from boreal forests to mixed deciduous/boreal forests. The shared morphological characteristics between Hi-Lo and Dalton result from a common antecedent, cluster, possibly southern Quad—a transitional Paleoindian/Archaic point type named after the Quad site in Alabama (Soday 1954).
Deller (Ellis and Deller 2014: 1) also remains skeptical of suggestions of continuity between
Holcombe and Hi-Lo.
2.4 The Hi-Lo Phase
The Hi-Lo projectile point type was first reported by James Fitting (1963) who named the site, and consequently the point type, after the Hi-Lo Gun Club in southern Michigan.
16
2.4.A Type Description
Based on a number of shared toolkit characteristics including; alternate beveling of point fore- sections, projectile hafting modifications and tool recycle forms, Hi-Lo has come to be seen as a regional manifestation of the Dalton horizon (see Tuck 1974; Koldehoff and Loebel 2009) found in areas of the United States south of the Great Lakes region (Shay 1971: 70; Ellis and Deller
1982, 17; Dickson 2011). Stylistic similarity and non-overlapping distributions have been seen as evidence for contemporaneity between Hi-Lo and Dalton (White 2013). Goodyear (1982) proposed an age range of 10,300 – 9,900 BP for the Dalton industry. Goodyear’s age range for
Dalton and by extension Hi-Lo is generally accepted by researchers (Koldehoff and Loebel 2009:
138-139; Jennings 2010; White 2013; Ellis et al. 2011: 536).
Hi-Lo points typically display excurvate blades with concave bases. Points are noted for their concave base with “eared" haft elements. Hi-Lo points often exhibit basal and lateral grinding which may in some cases distinguish the blade from the hafting element. Shouldering may occur at the juncture of blade and haft element. In extensively resharpened points shoulders have been reduced to a “nubbin” (Ellis and Deller 1982: 7) or removed altogether. Points with little evidence of resharpening tend to have excurvate blade edges. The effects of resharpening may remove or reduce blade curvature. Resharpened points may have straight to incurvate lateral blade edges. Hi-
Lo blade shape is extremely variable due to the effects of resharpening. Some have inferred the high degree of variability among Hi-Lo points to be a result of differing life histories of the individual points (Ellis and Deller 1982: 7). Analysis of blade morphology suggests that some points may have alternatively functioned as drills, side-scrapers or knives which heavily influenced blade shape (Ellis and Deller 1982: 7, 8).
A typical Hi-Lo point displays maximum width and thickness at the blade midpoint.
Unresharpened Hi-Lo points may show a well-executed collateral flaking or roughly parallel removal patterns. Resharpened points do not show consistent removal strategy and flake scars are
17 more randomly distributed (Ellis 2004a). Highly resharpened points show maximum width and thickness at the shoulders or at top of lateral grinding. Biconvex to plano-convex cross-sections are common among unresharpened points. Resharpened points tend to display a biconvex to
"twisted" parallelogram cross-section. The twisted orientation of a point’s cross-section is due to alternate edge beveling. Beveling (Andrevsky 2005: 253) refers to the process of modifying a tool edge to produce a desired edge angle through the removal of a series of flakes. Beveling was achieved through bifacial reduction rather than the unifacial beveling which is typically found on
Dalton points (Ellis and Deller 1982: 7).
Hi-Lo appears to exist at a transitional phase between the late Paleoindian period and early
Archaic. The term Archaic is used to describe “artifact forms and site types, rather than in much- debated inferred sociocultural terms” (Ellis et al. 2009: 788). The division between Paleoindian period and the succeeding Archaic is understood here to be “… marked on the early end by certain assemblages dominated by notched or markedly stemmed point forms dating to around
10,000.” (Ellis et al. 2009: 788). Hi-Lo has been regarded as Paleoindian (Ellis and Deller 1982;
Fitting 1963; White 2013: 84), early Archaic (Wright 1995: 71-73), or even both (Ellis 2004a;
Ellis and Deller 1990: 57-58; Ellis et al. 1990: 71). While Hi-Lo production shares almost all of its characteristics and traits with the preceding Paleoindian point forms, it differs in morphology and use (Ellis 2004a, 77).
Ellis and Deller’s (1982, 2014) studies of Hi-Lo points from southwestern Ontario comprise the most comprehensive review of the type in Ontario.
2.4.B Hi-Lo Subtypes
Documenting and classifying the variation within the Hi-Lo type is an enduring avenue of research. Fitting (1963) attempted to identify subtypes within Hi-Lo based on variation in blade length but later came to understand variability in blade length to be a function of resharpening
18 modifications affecting the blade’s length. As such, his proposed subtypes were realized to have little use as markers showing stylistic change over time (Fitting 1970: 42).
More recently, Ellis (2004a) has divided Hi-Lo into three subtypes based upon differences in hafting element configuration. However, at this time Hi-Lo largely continues to be reported as a monolithic type within the literature.
Figure 3. Hi-Lo subtypes proposed by Ellis 2004a
(1) The proposed earliest of the subtypes is the Holcombe-like Hi-Lo or “Hi-Ho” point form
(see Jackson 1998; Timmermans 1999). This form is considered earliest based on its similarity to
Holcombe forms (Fitting et al. 1966). Hi-Ho points have a lanceolate form and lack any distinctive hafting element (i.e. stem or side notching). They are less finely flaked and notably thicker than ultra-thin and carefully flaked Holcombe points (Ellis 2004a: 64)
(2) The “Classic” Hi-Lo point has a slight stem with slightly concave lateral stem margins.
Expansion beyond the stem is often largely removed over the course of the tool’s use-life through resharpening of the lateral fore-section. This results in small “nubbins” representing the remnants of the original stem-fore section juncture. Fore section resharpening is most often achieved
19 though alternate edge beveling. The “Classic” Hi-Lo subtype is the most common Hi-Lo form recovered from sites such as Welke-Tonkonoh, Stewart and ENL 502 (Ellis 2004a: 63; Roberts
1985).
(3) The proposed latest subtype is the side-notched Hi-Lo. This form bears a strong similarity to the “Classic” stemmed Hi-Lo but instead exhibits shallow side-notches. Hi-Lo points fitting this description has been recovered from sites such as Stephenson, Stelco I and Ageing Maple.
Side-notched Hi-Lo points also been recovered from Stewart site and the Welke-Tonkonoh sites
(Ellis 2004b) which have also yielded “Classic” stemmed Hi-Lo points. Notched Hi-Lo points are seen as the latest in this temporal sequence as notching is seen as a characteristic of the early
Archaic rather than the late Paleoindian (Ellis et al. 2009: 791).
2.4.C Hi-Lo Raw Material Usage Patterns in Ontario
Hi-Lo lithic raw material usage patterns have received substantially more attention than other late
Paleoindian technological systems. Haldimand and Kettle Point chert are the predominant source materials for Hi-Lo points reported in the Ontario literature. After examining a sample (n=102) of
Hi-Lo points from southwestern Ontario, Ellis and Deller (1982: 7) report Haldimand chert as the most popular material, comprising 38.24% (n= 39) of points available for study. The frequency of
Haldimand chert is almost equaled by Kettle Point chert at 31.37% (n= 32). Outside of these two main sources there are a few other less popular types of toolstone known to be used in Hi-Lo manufacture. Onondaga chert was identified as the material for 6.68% (n= 7) of the sample and
Bayport chert from Michigan on 7.84 (n=8). The remaining 15.69% (n=16) of points included in the study were made from an unidentifiable toolstone.
Examination of original unflaked portions of Hi-Lo points displaying cortex suggests that primary chert outcrops were utilized rather than secondary sources such as river cobbles or chert mixed in with glacial till (Ellis 2004a: 60). Parker (1986a) has found evidence of a Hi-Lo occupation at the
20 site of a Haldimand chert outcrop. The preference for primary chert outcrops rather than secondary deposits represents a continuity with early Paleoindian raw material procurement patterns. It stands in opposition to subsequent early Archaic raw material procurement patterns which show substantial use of secondary deposits (see Ellis et al. 2009).
Hi-Lo groups show a strong preference for Kettle Point or Haldimand cherts. Hi-Lo lithic assemblages have been found over 125km away from a chert outcrop yet may be composed of
80% of said chert (Ellis 2004a: 61). This pattern holds true for both the Welke-Tonkonoh (AfHj-
5) and the ENL 502 (AlGo-36), both of which yielded assemblages of over 80% Haldimand chert despite being 175 and 120km away from the primary outcrop (Parker 1986a; Roberts 1985; Ellis
2004b). Ellis (2004a: 61) disputes Roberts’ (1985: 211-212) raw material identification for the
ENL 502 Hi-Lo site. Roberts identified Ancaster chert as the source material for the site’s assemblage but Ellis believes that this is due to Roberts being unaware of the existence of
Haldimand chert. Roberts’ identification occurred before major developments in Ontario chert sourcing (i.e. Parker 1986a; Fox 2009). Ellis and Deller (2002) also report a cache of tools from the Caradoc site (AfHj -104) associated with a Hi-Ho occupation made of Bayport chert from over 200km away in Michigan. These assemblages point toward a pattern of little distance decay for Hi-Lo lithic assemblages (Ellis 2004a: 61).
Haldimand chert outcrops on top of the Onondaga Escarpment in the western Niagara region
(Moerschfelder 1985: 8) while Kettle Point chert occurs on the shores Lake Huron at Kettle Point,
Ontario (Fox 2009: 362). These cherts outcrop at the western and eastern extremes of southern
Ontario. The predominant usage of these cherts in place of northerly Collingwood chert suggests a change in mobility patterns. Earlier Paleoindian groups have been demonstrated to largely follow a north-south movement pattern but the same cannot be said about Hi-Lo groups. North- south mobility patterns seem to have terminated along with the use of Collingwood chert. Hi-Lo groups must have had an east-west mobility orientation in order to enable the exploitation both
21
Kettle Point and Haldimand chert. This interpretation assumes direct acquisition of toolstone as embedded within Hi-Lo movement. In fact, a side-notched Hi-Lo point was recovered at a
Haldimand chert quarry site (Parker 1986a; 1986b).
Hi-Lo raw material procurement patterns are notable for the avoidance of chert types common among assemblages over the course of Ontario’s prehistory. Hi-Lo raw material preference for primarily Haldimand and Kettle Point cherts appears to be at the expense of other readily available materials. Onondaga chert, identified as the raw material of 6.68% (n= 7) of the points included in Ellis and Deller’s (1982) sample, is notable for its scarcity in Hi-Lo assemblages.
Onondaga chert was perhaps the most popular chert for Aboriginal flint-knappers in southern
Ontario during prehistory (Fox 2009: 361) and its avoidance by the peoples of the Hi-Lo Phase is conspicuous.
Onondaga chert outcrops in very close proximity to Haldimand chert as do a number of other chert varieties. It is considered within Ontario to be a high quality chert due to the thickness of the bedding, up to 13cm (Fox 2009: 361), and its workability. Onondaga is considered to be of higher quality than Haldimand chert and thus the preference of Hi-Lo peoples for Haldimand chert cannot be explained by a desire for higher quality material (Ellis 2004a: 61).
The apparent preference for Haldimand or Kettle Point cherts is at odds with the lithic procurement practices of earlier Paleoindian peoples (Deller 1988, 1989). Chert preferences can be seen as shifting towards the use of local materials as opposed to the preferred use of distant cherts by earlier Paleoindian groups. A poignant example is the Hi-Lo assemblage at the Stelco I site (AfHa-200). The Stelco I site at the mouth of the Grand River, excavated by Timmins (1995), has yielded an assemblages of 242 pieces of debitage and 38 tool forms. The assemblage is composed of only 1% Onondaga chert despite being located 8km closer to Onondaga outcrops than the Haldimand chert. With the exception of a single piece of Onondaga flaking debris, the entire Stelco I lithic assemblage is composed of Haldimand chert. Onondaga chert outcrops closer
22 to Stelco I than Haldimand (13 vs. 21km) and its paucity is of equal note to Haldimand chert’s dominance.
Early Paleoindians relied almost exclusively on Collingwood and Onondaga cherts while ignoring
Haldimand and the inaccessible Kettle Point bedrock outcrop (Deller 1989; Deller and Ellis 1988:
252). Since Onondaga chert was used heavily in the early Paleoindian period (see Ellis and Deller
1990) it is difficult to make a case for Hi-Lo groups being ignorant of its existence. Ellis has argued elsewhere (Ellis 1984, 1989) that Paleoindian raw material procurement patterns cannot be fully explained by utilitarian reasons and the frequency of lower quality materials must be the result of non-utilitarian factors influencing raw material selection.
The Welke-Tonkonoh site (AfHj-5) was investigated by Roosa in the 1970’s and once more in the
1980s by Ellis through a surface collection and partial excavation (Ellis 1981b, 2004b; Ellis and
Deller 1982). Welke-Tonkonoh remains the most extensive Hi-Lo site known with five discrete artifact scatters and a single component area. Area C of the Welke-Tonkonoh site has yielded a tool assemblage including a two complete side-notched Hi-Lo points and two foresections (Ellis
2004b). The singular presence of side-notched Hi-Lo points could support early Archaic placement though Ellis (1981b: 16) notes “the assemblage is very similar morphologically and technologically to earlier assemblages, particularly those associated with fluted points” The majority of the tools (n= 27) have been manufactured from Haldimand chert from primary outcrops 175km to the southeast of the site (Ellis 2004b).
Flaking debris at Area C was highly concentrated in an area of about 16 square metres. 17 pieces of debitage from three different sources were recovered; Haldimand chert (n= 7), Onondaga chert
(n= 6) and Kettle Point chert (n= 4). With the exception of two large pieces of Kettle Point chert, none of the lithic debitage weighed more than 0.68g, averaging under 0.3g. The assemblage was largely composed of bifacial thinning flakes or flake fragments. The debitage from Area C has been interpreted as resulting mainly from the final stages of tool production or resharpening. The
23 high ratio of tool-debitage in Area C is consistent with expectations for Paleoindian sites at a great distance from the assemblage’s chert sources (Ellis 2004b). Area C and the ENL Hi-Lo site
(Roberts 1985) are the only single component Hi-Lo sites known in Ontario (Ellis 2004b).
2.4.D Distribution and Excavated Sites
Hi-Lo is the most common late Paleoindian point found in southern Ontario (Deller 1989).
However, most Ontario Hi-Lo sites prominent in academic literature are only known through surface collection (e.g. Bursey 1998; Timmermans 1999). Hi-Lo points in Ontario are most often discovered as isolated surface finds. Only twelve Hi-Lo sites have been excavated in the province
(Table 2).
Table 2. Hi-Lo sites excavated in Ontario
Site Name Borden # Reference Stewart AfHj-6 Deller 1988b, 284-288; Ellis and Deller 1982 Strathroy AfHj-7 Deller 1988b, 284-288; Ellis and Deller 1982 Allan AfGx-50 Parker 1986a, 1986b Stelco I AfHa-200 Timmins 1995 Caradoc AfHj-104 Deller and Ellis 2001 Ageing Maple AkGv-91 Murray 1997 Witz AhHa-80 Murray 1997 Welke-Tonkonoh AfHj-5 Ellis and Deller 1982; Ellis 2004a, 2004b Snowhill Site AgHb-239 TMHC 2004 Double Take AgHb-240 TMHC 2004; Dickson 2011 Koeppe II AhHa-157 Woodley 1997 ENL 502 AlGo-36 Roberts 1985
Recently, the Double Take site (AgHb-240) (TMHC 2004) assemblage has been studied by
Dickson (2011) with an emphasis on toolkit composition. Koeppe II (AhHa-157), was test excavated and although no diagnostic points were recovered the existence of a Hi-Lo component is probable (Woodley 1997).
Hi-Lo points are mainly found throughout in the lower Great Lakes region but the distribution may reach a far west as eastern Wisconsin (Ellis et al., 2011; Justice 1987). Within Ontario, Hi-Lo
24 occurs only in the southern portion of the province (Ellis 2004a). To the east, Hi-Lo points are not known to occur in the New England or Maritime regions (Bradley et al. 2008).
2.5 Great Lakes Paleoindian Technology
In most cases, only the stone implements of the Paleoindian toolkit have preserved for archaeologists today. The study of these tool forms has been used to provide a chronology of the period and also to make inferences about Paleoindian lifeways. The purpose of this section is to review Paleoindian raw material usage patterns and explore aspects of the Paleoindian toolkit beyond the diagnostic biface types.
In order to properly interpret Paleoindian assemblages it is necessary to review the principle toolstone utilized during the period. The following section is meant to introduce the reader to the chert types most common in Ontario Paleoindian assemblages.
Chert is by and far the most common lithic material to be used in prehistoric Ontario. Chert is a variety of the naturally occurring mineral quartz. It is composed of silica (silicone dioxide SiO2) and various impurities including carbonate, clay, carbon, iron or manganese oxides, and iron sulphide. It has a crystalline form of quartz composed of microcrystalline and cryptocrystalline crystals. The difference serves to distinguish chert from other silica based material such as obsidian which are non-crystalline (Eley and Von Bitter 1989). Following Eley and Von Bitter
(1989: v) “chert,” is used to refer to “all the bedded and nodular deposits in sedimentary rocks that consist of cryptocrystalline, microcrystalline, or microfibrous forms of quartz.” In order to differentiate between varieties of chert archaeologists have frequently chosen to assign geographic names to chert types in order to differentiate between them (Eley and Von Bitter
1989).
25
2.6 Raw Materials Common in Ontario Paleoindian Assemblages
(1) Bayport chert occurring in the Upper Mississippian Bayport formation of the Saginaw
Bay area of Michigan. Bayport chert may occur in bedded forms but nodular forms appear to more common. Bayport chert may gradually blend into the surrounding limestone matrix or an abrupt division may separate the chert from its matrix. The chert is light grey to brownish grey to dark greenish grey in colour. It is opaque with a dull lustre, it is coarse to medium gained in texture and may have some small quartz inclusions. In some cases iron oxides are present on the surface though this is not always the case (Ellis and Deller 2000). A diagnostic characteristic of
Bayport chert is concentric rings originating at the centre of the larger nodules. The concentric banding is not known on any Ontario cherts and may serve to distinguish Bayport chert. A notable attribute of Bayport chert is its characteristic weathering pattern. Bayport chert often develops a brownish patina which will be most commonly present on specimens of great antiquity (Ellis and
Deller 2000: 43)
(2) The Collingwood chert occurs as part of the larger Middle Silurian Fossil Hill formation within the Niagara Escarpment (Fox 2009: 360). Chert of the Fossil Hill formation also outcrops on Manitoulin Island though this material is easily distinguishable from the chert located in the
Beaver Valley near Collingwood (Deller and Ellis 2011). Collingwood chert occurs as secondary deposits south of the area of the primary outcrops. Collingwood chert outcrops in beds that can reach up to 20cm in thickness (Deller and Ellis 2011). Collingwood chert beds often terminated in an abrupt and regular juncture with the dolomitic cortex. The chert ranges in colour from white to light grey to a very brownish beige. Reddish brown staining may occur due to the presence of iron oxides. It is relatively fine grained and ranges from opaque to slightly translucent. It is considered to have a medium to high luster. Banding may occur running parallel to the juncture with the chert’s cortex (Deller and Ellis 2011).
26
(3) Haldimand chert is found in the Bois Blanc formation of southern Ontario which also yields Saugeen and Colborne cherts (Parker 1986a, 1986b). The Bois Blanc formation was formed during the Lower Devonian period over 350 million years ago (Fox 1978: 6; Telford and
Tarrant 1975). The formation underlies the Onondaga Escarpment in the county of Haldimand, north of Lake Erie. The chert itself can be observed as discontinuous beds, or as nodules, in the bedrock sections of the many quarries between Cayuga and Hagersville, Ontario. Haldimand chert varies considerably in lustre, and ranges in colour from white, light gray, blue-gray to pink.
It occurs in the uppermost portions of the Bois Blanc formation, and can reach thicknesses of 70 mm or more (Fox 1978: 6).
(4) The Kettle Point chert is the youngest chert type in Ontario. It is found in the upper
Middle Devonian Ipperwash formation of the Hamilton Group. Kettle Point chert is intercalated with limestone and occurs in a very thick bed, ranging from 2-20cm, at the top of the Ipperwash limestone (Eley and Von Bitter 1989: 15). The only known natural outcrop of Kettle Point chert occurs on the shores Lake Huron at Kettle Point, Ontario. Massive siliceous beds characterize the outcrops at Kettle Point, where it was actively quarried by Aboriginal groups for millennia (Fox
2009: 362). Kettle Point is a fine grained chert with a waxy lustre. It is often occurs in laminated bed cross cut by quartz and limestone (Eley and Von Bitter 1989: 15). Kettle Point chert ranges from reddish brown to blueish gray to black (Eley and Von Bitter 1989: 15; Janusas 1984: 2-5) It is important to note that the Kettle Point outcrop was submerged during the early Paleoindian period (Deller 1989). Its inaccessibility at this time was a result of elevated water levels in the
Huron basin due to the presence of glacial Lake Algonquin. The inaccessibility of the primary outcrop does not necessarily mean that Kettle Point chert was unknown to peoples of the early
Paleoindian period (Ellis and Deller 2000: 184). Secondary deposits occur south of the primary outcrop through glacial transport.
27
(5) The Onondaga formation consists of three members; the Edgecliff, Clarence, and
Moorehouse. The Edgecliff member is the major producer utilized chert in the Onondaga formation (Eley and Von Bitter 1989: Fig. 1). Onondaga chert primarily occurs in the Clarence member of the lower Middle Devonian Onondaga formation (Parkins 1977). All three members of the Onondaga formation produce chert in the form of primary outcrops. These outcrops occur along or near the current Lake Erie shoreline in Ontario and northwest to the village of Villa Nova and extend eastward into New York State. Onondaga chert is a mottled, fine grained chert. It varies from dark blueish grey to a light grey mottled colour. It is of medium lustre, opaque and often includes quartz inclusions and occasional iron oxide. Onondaga chert may blend into the surrounding limestone matrix, junctures are highly irregular (Ellis and Deller 2000: 42).
Onondaga chert is known to occur exist in secondary deposits throughout southwestern Ontario.
Glacially transported blocks of Onondaga chert have been noted by Fox (1989: 26) as far west as
Pelee Island.
2.7 Early Paleoindian Raw Material Usage Patterns in Ontario
2.7.A Gainey Phase
Gainey phase lithics are exclusively made by chipping or flaking. They are characteristically made of high grade cryptocrystalline chert originating great distances from site locations
(Goodyear 1989; Melter 1984; Ellis and Deller 1990). Upper Mercer chert from Ohio is very rare in Ontario and during the Paleoindian period it was only utilized by Gainey phase peoples
(Jackson 1996). The Paleoindian use of Upper Mercer chert terminates with the end of the Gainey phase (Deller 1989).
Some Gainey points in Ontario are manufactured on Collingwood chert (Ellis 1984a) however, most are made from Onondaga or distant cherts such as Upper Mercer, Flint Ridge and Tenmile
Creek chert from Ohio (Ellis 1984a). The Udora assemblage is composed of mainly Collingwood
28 chert although significant amounts of bedrock derived Onondaga chert from approximately 120 km to the southwest and local Balsam Lake chert are also present (Storck and Speiss 1994)
2.7.B Parkhill Phase
Parkhill and Fisher are the largest Parkhill phase sites in Ontario. The Parkhill site is positioned on a strandline of what would have been the main stage of glacial Lake Algonquin (Karrow et al.
1975). With only a single exception, the lithic tools at the Parkhill site are composed of three types of chert: the Collingwood, Onondaga and Bayport chert (Deller and Ellis 1992b; Ellis and
Deller 2000). Collingwood chert is by far the most common among of Parkhill lithics; it comprises 86.4% of the total assemblage while Onondaga and Bayport chert comprise 6.1% and
6.4% respectively (Ellis and Deller 2000: Table 4.1). The limited Kettle Point chert present in the
Parkhill assemblage have weathered surfaces indicating that they were likely obtained from secondary deposits (Ellis and Deller 2000: 184).
Deller and Ellis (1992a) argue that Parkhill peoples were able to achieve a much greater consistency in blank production than earlier Gainey peoples. Blanks are specific detached pieces of chert capable of modification into further tool form (Andrefskey 2005: 253). Blanks were created by striking large blocks of chert derived from primary outcrops in order to cleave off a long flake of chert running perpendicular to the cortex of the block. The banding on Collingwood chert, which is often found running parallel to the cortical edges the block, is consistently oriented perpendicular to the longitudal axis of Barnes points (Deller and Ellis 1992a: 31). A consistent blank production strategy allowed for a greater standardization of points and more efficient use of raw materials. At the Parkhill site, 44 of 56 fluted points were initially formed as blanks in this way and exhibit the same perpendicular banding-point orientation. In addition, at the Parkhill phase Thedford II site, 12 of the 15 points made from Collingwood chert were manufactured using an identical blank production method (Deller and Ellis 1992a). This finding supports a
29 widespread Barnes point production strategy and consistent approach to primary block reduction among Parkhill phase sites
The Fisher site is a large Parkhill phase site located in the southern Georgian Bay area of south- central Ontario. The site occupies the highest ground in the region which remains in close proximity to the Lake Algonquin. The Fisher site is composed of a total of 19 discrete artifact clusters. It was first discovered in 1975 as a part of a long-term ROM research initiative meant to investigate the age, technology and cultural adaptations of early Paleoindians in Ontario (Storck
1974a, 1974b, 1979, 1982, 1984a, 1984b; Storck and Speiss 1994). The site is single component and can thus offer insights into the Parkhill Phase without any later occupations obscuring the record (Storck 1997: 8).
The Fisher site presents an alternate approach to fluted point manufacture. A consistent blank production technique noted at Parkhill and Thedford II is not readily apparent at the Fisher site
(Storck 1997: Fig. 8.2). Banding, when identifiable, displays a broad range of orientations. This may indicate a less systematic approach to point manufacture than that employed by the groups who occupied the Parkhill and Thedford II sites. The greater variation in banding-point orientation found among the points from the Fisher site may also be a result of the effects of a larger sample size. In separate analyses Storck identified 71 points displaying banding (1997: Fig
8.2) while Ellis noted 96 (Deller and Ellis 1992a: Fig. 41). A large collection of points may be expected to show greater variation than a smaller assemblage such as that of Thedford II.
2.7.C Crowfield Phase
Crowfield phase sites are very rare in Ontario, and as such, it is difficult to make inferences about
Crowfield raw material usage patterns. The Crowfield phase shows heavy use of Onondaga chert with minimal use of Collingwood or Bayport chert (Jackson 1998: 12). The Bolton site assemblage was manufactured primarily on bedrock derived Onondaga chert originating some
30
100 km southeast. This site shows a clear preference for Onondaga chert in contrast to the predominant Collingwood chert usage characteristic of the Parkhill phase. (Deller and Ellis 1996).
2.8 Late Paleoindian Raw Material Usage Patterns in Ontario (Excluding Hi-Lo)
2.8.A Holcombe Phase
Holcombe points have been found across the Great Lakes region manufactured from
Collingwood, Bayport, Tenmile Creek, Onondaga, Upper Mercer and Kettle Point cherts (Ellis
1986). Within Ontario, the use of Kettle Point and Onondaga chert as the source material for
Holcombe points is typical (Jackson 2004), however Collingwood specimens exist (Woodley
2004). At the Holcombe type site, Fitting et al. (1966: 126) recorded the almost exclusive use of
Bayport chert from outcrops located approximately 160-180 km north of the site location.
The Fowler site is the only single component Holcombe site which has been excavated in Ontario
(Woodley 2004). The lithic assemblage recovered from Fowler included 2,105 pieces of debitage and five different raw material types. The majority of the assemblage was comprised of Onondaga chert (87%) or Collingwood (12%) chert, while Kettle Point chert was also present in negligible amounts. In addition, quartzite and greywacke tools were also recovered. Nine Holcombe projectile points were recovered from the site (Woodley 2004).
2.8.B Madina Phase
Madina phase lithic material usage patterns are poorly understood in Ontario. At this time points are primarily known in the northern portion of southern Ontario. Point distribution is highly restricted which hinders out ability to see larger usage patterns. Madina points have been known to occur on Onondaga, Collingwood, Balsam Lake, Trent, and Kettle Point cherts along with other unidentified exotic cherts (Dibb and Ellis 1988: 20)
31
Madina phase lithic procurement may show evidence for “settling in”. Madina points manufactured on Trent, Collingwood and Onondaga cherts hint at a more generalized toolstone procurement pattern (Dibb 2004: 156). 14 different types of lithic raw material were recovered from late Paleoindian components in the Queensville-Keswick area (Dibb 2004: 131). While the method of procurement for these materials is not yet known it is of note that such a great variety of materials are present. The assemblage includes many cherts which were not utilized during earlier times. The most common material noted is Balsam Lake chert which belongs to the upper member of the Bobcaygeon Formation which overlays the Gull River Formation. Balsam Lake chert outcrops at points along the shore of its namesake as well as appearing as nodules in nearby lakes (Eley and Von Bitter 1989). The use of more local material stands at odds with early
Ontario Paleoindian preferences for Collingwood chert or less frequently, Onondaga chert. The use of these more local cherts also points towards a shift away from the use of high quality cryptocrystalline raw materials seen in earlier times (Goodyear 1989; Haynes 1980: 118; Meltzer
1984).
2.9 Paleoindian Mobility
Paleoindians have traditionally been seen as small, highly mobile groups exploiting large annual ranges (e.g., Goodyear 1989; Kelly and Todd 1988). The mobility of these groups is often thought to be in excess of ethnographically known groups (e.g. Amick 1996: 419–420; Ellis 2011; Kelly and Todd 1988; Shott 1986: 141–142; Storck and Tomenchuk 1990: 84). Criticism of these estimates has often been dismissed as over reliant on the ethnographic record (Speth et al. 2011:
1). Estimates of Paleoindian mobility are often linked to subsistence strategies and raw material procurement patterns. Changes to Paleoindian mobility patterns have been closely linked to changes in subsistence strategy. Late Paleoindian like Hi-Lo are seen as having begun to settling in to their landscape and exploiting a wider range of resources than early Paleoindians.
32
2.9.A The High Technology Forager Model
The Kelly and Todd’s (1988) conception of early Paleoindian lifeways draws on data from across
North America and particularly the Great Plains. Interpretation of early Paleoindian sites across
North America are often influenced by the “high technology forager” model. In the model, the presence of exotic toolstone in assemblages is considered to be evidence for high residential mobility among Paleoindian groups. The lack of reuse of kill sites is thought to be the result of unpredictable land use patterns as groups frequently shifted their range. Paleoindian technology is interpreted as designed to economize raw material. Early Paleoindians are thought to have colonized an unpopulated or sparsely populated landscape. With little or no prior knowledge of the landscape and quarry locations, toolstone had to be carefully conserved and efficiently utilized. The Paleoindian toolkit was designed to be efficient and multipurpose in order to reduce overall carrying weight. As a result of their high mobility across different ecosystems the
Paleoindian toolkit in a high technology forager model should display limited technological variation (Kelly and Todd 1988: 235).
Paleoindians lived at a when North America featured environments with no modern analogues, therefore conceptions of Paleoindian lifeways need not to be constrained by the ethnographic record (Wobst 1977). Early Paleoindian assemblages across the continent are typified by the prominent use of well-made bifaces, high-quality cryptocrystalline raw materials, and tools with the potential for long use-life, most often projectile points i.e. Clovis or Clovis-like Gainey points
(Goodyear 1989). These characteristics have been interpreted as an indication of a Paleoindian lithic technology which was meant to be highly transportable with long-term utility, marked by the stringent use of raw material to conserve resources in areas where few toolstone sources were known (Kelly and Todd 1988: 237-238). A high technology forager model sees early Paleoindians as technology oriented rather than place oriented meaning that early Paleoindian technology was
33 not developed to respond to immediate environmental conditions but rather meant to allow its users to respond to unfamiliar landscapes and conditions.
A diet based primarily on large terrestrial game rather than floral resources is posited by Kelly and Todd (1988) as an aid to Paleoindian colonization of unknown areas. Large game provide an easily procurable, widespread and year-round source of food. These animals are highly visible and, in an unfamiliar environment, may provide a food source that is easier to locate, procure and process than floral resources whose distribution and fruiting schedules may be unknown.
Following the annual rounds of migratory animals such as bison or caribou would enable
Paleoindians to traverse regional environmental boundaries without the dangers associated with subsisting in an alien environment (Kelly and Todd 1988: 234)
Kelly (1995: 131-132) has argued that the possibility of obtaining resources with higher return rates, namely large game, incentivizes and enables long distance logistical mobility. In a similar way, Paleoindians can be envisioned as highly residentially mobile as they move to exploit distant resources, whether to acquire toolstone or for subsistence purposes. The exploitation of migratory caribou during the course of their seasonal migrations is one of the possible strategy that would allow for such a residential mobility pattern (Ellis 2011: 398). In an open landscape with low population density, groups would be able to target resources they viewed as attractive without major competition from other groups (Speth et al. 2011). For early Paleoindians first colonizing a recently de-glaciated Ontario, conditions allowing for extraordinary mobility and selective targeting of high value resources (such a migrating caribou) would almost certainly have existed
(see Ellis 2011: 394).
34
2.9.B Northeastern Paleoindian Mobility and Caribou Predation
In the Northeast, Paleoindian mobility has been frequently tied to the exploitation of caribou (e.g.,
Curran and Grimes 1989: 59–60; Koldehoff and Loebel 2009: 282; Meltzer 1988: 41; Newby et al. 2005: 150–151; Spiess et al. 1998). A group’s annual range mobility (Binford 1983: 38) has been frequently assumed by researchers to be represented by the presence of exotic toolstone in an assemblage (Ellis 2011). Assuming that this toolstone was acquired by the group at a primary outcrop, exotic toolstones may provide some measure of annual range mobility. Northeastern early Paleoindians peoples are thought to have been highly mobile, to such a degree that estimates of their mobility range are unusually large compared to historically known foraging societies documented in the ethnographic record (Ellis 2011: 386).
2.9.C Early Paleoindian Raw Material Procurement and Mobility
Range mobility estimates based on the presence of exotic lithic material may not necessarily be representative of the total movement of a people. The exploitation of a chert outcrop of known provenience cannot confidently be placed within a group’s movement pattern. Even though toolstone acquisition may be embedded in a group’s mobility pattern it cannot be said whether acquisition exists at the extreme of the group’s range, in the middle of the seasonal round or outside of the group’s normal mobility (Storck and Von Bitter 1989). The idea of direct acquisition of toolstone is an important tenet of many Paleoindian mobility estimates. Goodyear
(1989) sees the presence of distant materials in site assemblages as strong evidence for the mobility of Paleoindian groups. The high degree of mobility for Paleoindians was made possible by the use of these high quality materials, “portable technologies” (Goodyear 1989) and careful curation of tools to extract maximum utility.
While hunter-gatherers rarely move in direct straight lines, it is worth noting that the straight-line distance between Paleoindian sites and toolstone sources frequently exceed 250 km one way or
35
500 km for a round trip (Ellis 2011: Fig. 2). It is difficult to imagine a small band of hunter- gatherers thoroughly exploiting an area this size over the course of a yearly round. Ellis (2011:
392) proposes that Paleoindians would only exploit “… certain restricted locations on the landscape, then travel long distances to target resources in other locations, ignoring or only briefly exploiting much of the area in between.”
The assumption of embedded acquisition is taken as a given by many researchers but has not found support among all (Speth et al. 2011; Curran and Grimes 1989: 72; Deller 1989: 219;
Tankersley 1989: 270). Toolstone acquisition may have been a task delegated to logistical task groups and existed outside of the group’s normal mobility range (e.g. Amick et al. 1997: 170;
Spiess and Wilson 1989: 95–96). It is also possible that exotic cherts within an assemblage may have resulted from the movement of individuals from one band to another.
Early Paleoindian lithic assemblages often include lithic material which originated extraordinary distances from the site. This material may be from distances of 300 km to over 800 km away
(Haynes 1982: 392; Simons et al. 1984: 267; Tankersley 1989: 268). Paleoindian researchers often assert that chert was acquired through direct acquisition at the source and then carried from site to site. Toolstone may also be acquired though indirect acquisition. This occurs when a group acquires the material from the source and then transfers it to another group through some sort of exchange mechanism (Meltzer 1989).
Complicating matters further is the idea that exotic toolstone frequently seen as evidence of group residential mobility may be acquired though trade (Deller 1989; Ellis and Deller 1990: 54).
Storck and Von Bitter (1989) do not believe trade to be the source the toolstone required for the majority of the Paleoindian toolkit as trade would be a far too unreliable source for hunter- gatherers. They cite the fact that at distances of 175-200 km from the outcrop of Collingwood chert, its frequency in Parkhill sites lithic assemblages only falls from 95% to 85% (Storck and
Von Bitter 1989). In addition, at the Kolopore site (Storck 1982, 1984a) located in the area of the
36 outcrop there exists abundant evidence of primary block reduction and the preparation of bifaces but no evidence of finished tool forms. Thus, direct acquisition from the primary outcrop is hypothesized to be the preferred method of acquisition. The lack of complete tool forms and evidence for primary reduction is interpreted as evidence for logistical task groups making special trips to gather raw materials (Storck and Von Bitter 1989). This suggests an efficient lithic reduction strategy minimizing carrying costs.
While trade may not have been the main form of toolstone acquisition for Paleoindian groups it may still have played some part in maintaining inter-group relations. Goodyear (1989) believes that exchange would likely involve the movement of “finished items – not assemblages…”This position is consistent with Storck and Von Bitter’s direct acquisition model but allows for some flexibility. Deller (1989) sees trade of exotic toolstone as a possible indicator of inter-group relationships. At the Parkhill phase Thedford II site the lithic assemblage is dominated by
Collingwood chert; 86% of lithics recovered are of Collingwood chert while Bayport chert from
Michigan accounts for 13% of the assemblage. Collingwood chert is the source material for 69% of fluted points while Bayport chert accounts for the remaining 31%. These data suggests that fluted points themselves may not be representative of group chert usage and exploitation as a whole. The frequent use of exotic chert for point manufacture is taken by Deller as an indication of their importance for exchange relationships (Deller 1989).
Paleoindian sites in the Northeast typically display an overwhelming predominance of a single chert type. Ellis (1989: Table 6.1) cites 25 Paleoindian sites from across the northeastern region whose assemblages are dominated by a single chert source. Ellis asserts that almost all sites in the
Great Lakes region “strongly suggest that only one or a very limited number of sources may have been regularly visited in the course of an annual round” (Ellis 1989: 144). Storck and Von Bitter
(1989) note exhausted tools from Ontario quarry sites are primarily made from Collingwood chert
37 rather than distant, exotic cherts. This evidence strongly suggests that Paleoindian groups made intentional trips to primary outcrops to replenish their toolstone supply.
The number of Parkhill sites displaying the same type of chert preferences suggest a widespread, shared mobility and raw material exploitation pattern. In contrast to the raw material usage patterns of later groups, there is little gradually diminishing use of a preferred chert as one moves away from the source. Instead, it is suspected that at a certain distance there exists an abrupt drop off in the use of preferred chert types. No major Parkhill sites are known in Ontario, Michigan or
New York where the primary source of toolstone in the assemblage originated from a source more than 200 km from the site (Deller and Ellis 1992: 49).
Within Ontario, Storck (1982) has suggest that the Fisher and Parkhill sites may represent two parts of a seasonal migration route. These sites are hypothesized to have been occupied at different times during the seasonal rounds of Parkhill phase peoples as they migrated from the southern Georgian Bay region, near outcrops of Collingwood chert, to the area of the southern
Huron basin. The presence of Collingwood chert and Bayport chert at the Parkhill site are used by
Deller and Ellis (1992) to suggest a north-south migration for the Parkhill phase people. Raw material analysis of the Parkhill site assemblage indicates that toolstone was quarried from outcrops rather than collected from secondary deposits. The majority of lithic material recovered from Parkhill and associated sites around the Thedford Embayment are comprised of
Collingwood chert. Sites have yielded substantial tool assemblages (13 to > 300 tools) of which over 80% are on Collingwood chert from 175-200 km to the northeast (Storck and von Bitter
1981). Given the northerly location of Collingwood chert outcrops in relation to the Parkhill site and the use of bedrock derived chert it is reasonable to infer a general north-south migration of
Parkhill peoples.
Although the Crowfield phase remains poorly understood in Ontario, some inferences concerning mobility may be drawn from raw material usage patterns. The Crowfield phase shows heavy use
38 of Onondaga chert with minimal use of Collingwood or Bayport chert (Jackson 1998: 12). At the
Crowfield type site the lithic assemblage was composed of 75.3% Onondaga, 15.9% Collingwood and 7.1% Ancaster chert (Ellis 2009). The Bolton site assemblage was manufactured primarily on bedrock derived Onondaga chert originating some 100 km southeast but contained one piece of
Collingwood chert. The Bolton site shows a clear preference for Onondaga chert in contrast to the predominant Collingwood chert usage characteristic of the Parkhill phase (Deller and Ellis 1996).
Assuming that sampling bias has not skewed the record, Crowfield toolstone preference suggests discontinuity with the Parkhill phase and perhaps a shift in raw material exploitation patterns.
To understand the directionality of raw material procurement trips, Ellis (2011) measured the direction from Paleoindian sites (excluding Hi-Lo) to the source of the toolstone most well represented in the site’s assemblage. The dominant toolstone in an assemblage was used as a proxy for annual range mobility. This procedure was followed to avoid the confounding effect of small amounts of toolstone possibly obtained by other means than direct acquisition. If acquisition through logistical task groups or exchange formed a major part of the group’s toolstone acquisition it was predicted that there would be no patterning in the direction between sites and the principal toolstone source (Ellis 2011: 389-390). Alternatively, if direct procurement embedded in normal residential movement patterns was the preferred method, one would expect patterning in direction between sites and the main toolstone source.
Directions from sites to toolstone sources was measured using a 360 degree scale. Any site within
45 degrees either side of true north was grouped as “north’’ while those falling 45 within either side of true east as ‘‘east.’’ Within the sample (n= 83) twice as many sites were oriented north– south (n= 44 or 63.7%), as opposed to east–west (n= 24 or 36.2%) a difference significant at the
.05 level (X2 = 5.232, df = 1, p = .022). The north-south relationship between site and toolstone source is more pronounced among sites where the toolstone source is over 200 km away from the site location (Ellis 2011: 390). Results suggests that logistical trips to procure toolstone would
39 likely only have occurred when groups were relatively close to the desired outcrop. These data support Storck’s interpretation of the Kolopore site located in the immediate area of the
Collingwood chert outcrops (Storck 1982, 1984a).
2.9.D Ontario Early Paleoindian Mobility and Caribou Predation
The stylistic commonalities of the Fisher and Parkhill site fluted points and the similar manufacture sequence has been interpreted by Storck and von Bitter (1981: 30) as evidence that both sites were occupied by the same group of people. This would provide evidence of a movement of at least 185km as the Parkhill phase people moved on what is speculated to be a seasonal basis.
Movement between these two sites has been tied to the exploitation of migrating caribou. The direction of this movement between Fisher and Parkhill would have paralleled that of the caribou on their own seasonal rounds. Parkhill peoples may have been exploiting caribou from at least one and possibly both ends of their seasonal migrations (Storck and von Bitter 1981: 35; Ellis 1984d:
389-390; Deller 1988: 176). Additionally, the presence of tool caches at Thedford II indicates the probable intended return of Parkhill peoples to the site.
Evidence from the Parkhill and Fisher sites points towards a north-south seasonal round for the
Parkhill peoples. Chert was likely obtained during the snow-free months of the year and then moved south towards the southern Huron basin to exploit the annual migration of caribou to the area of the Parkhill site. Deller (1979: 15) and Storck (Storck and von Bitter 1981: 35; Storck
1984: 13-15) observe that the area around the Collingwood, Ontario and the Fisher site is noted for heavy snowfall during the winter months. This snowfall would undoubtedly have been even more extreme during the late Pleistocene. With this in mind, it seems difficult to imagine areas nearest to the Collingwood chert outcrops having been occupied by Parkhill peoples during the
40 winter months. However, this line of reasoning has not been accepted by all researchers (see Ellis
2011).
2.10 Implications of the Settling In Hypothesis for Late Paleoindian Mobility and Raw Material Procurement
As time passed, early Paleoindian colonists would have become more familiar with the landscape they had now come to inhabit. Ellis’ (2011: 387) “Settling In” model is based upon postulated decreases in mobility among foragers during the late Paleoindian period. When the first
Paleoindian colonists entered Ontario they would have encountered a scarcely or completely uninhabited landscape. As time passed the colonists became more familiar with their new home and populations increased. Groups became more familiar with local resources and broadened their resource base to exploit a range of food products previously ignored (Ellis et al. 1990: 66). As a result mobility declined among foragers though it still remained substantial. This can be regarded as an ongoing process throughout the Paleoindian period (Ellis 2011: 394). As stated earlier, decreasing regional mobility should be represented in the archaeological record by increasing stylistic regionalization (Ellis 2011: 394; Smallwood 2013: 692).
Settling in processes have been associated with the increased biomass and increasingly closed
Holocene forests (See Karrow and Warner 1990). Holocene conditions may have caused population growth and reduced hunter-gatherer mobility. These changes encouraged territoriality and with it the development of regionalized group identities. Regional stylistic variability resulted as an effect of the development social boundaries as inhibitors of cultural transmission (Koldehoff and Walthall 2004, 2009; Koldehoff and Loebel 2009).
Stylistic regionalization has also been interpreted as a result of reductions in mobility following the colonization process. Familiarization with the landscape led to the establishment of areas of habitual use. As populations grew, social interaction mechanisms became established. The
41 appearance of regional design styles is a result of the formalization of social interaction mechanisms (Anderson 1995)
2.10.A Late Paleoindian Mobility and Caribou Predation
The shift from fluted to non-fluted points in the eastern North America is thought to represent an adaptive shift away from the specialized hunting of megafauna and ungulates (Anderson 1990:
18). It is thought that hunters shifted their focus towards smaller game such as deer (Goodyear
1974, Morse 1973: 30, 195-197; Smith 1986: 9-13). As foragers became less mobile, they would have become increasingly subject to local environmental conditions. Groups no longer traversed great distances across multiple ecosystems but became increasingly tethered to specific regions.
Increased regionalization encouraged the development of diverse subsistence strategies across space as hunter-gathers adapted to heterogeneous environments (Cannon and Meltzer 2008: 15).
However, a shift away from caribou predation during the late Paleoindian period is not assumed by all late Paleoindian researchers (e.g. Dibb 1985, 2004; Woodley 2004). In fact, caribou remains were recovered from the Holcombe site (Fitting et al.1966). Woodland caribou remained extant in southern Ontario into the modern era and would have been available for hunters
(Bergerud 1974). However, it is possible that range shifts occurred as a result of changing environmental conditions during the transition from the Pleistocene to the Holocene. If late
Paleoindian mobility patterns were in fact influenced by seasonal caribou exploitation, changes in range and herd condition are important considerations.
A change from early Paleoindian mobility patterns and caribou exploitation is not universally assumed for the Holcombe phase. Woodley (2004) envisions the Fowler site as a warm weather camp ideally situated to intercept migrating caribou. He observes that caribou tend to move further south during the colder month and therefore it is most likely that both herd and hunter would likely have moved further to the south during the winter months. This interpretation is
42 consistent with Fitting et al.’s (1966) interpretation of the Holcombe Beach site in Michigan as a warm weather camp. If this assumption is correct then the Holcombe peoples would be following an annual north-south mobility pattern similar to those seen in the early Paleoindian period (Ellis
2011: 386).
Given the severely restricted distribution of Madina phase materials very little can be said with regards to annual mobility. Madina points have also been recovered from the Heaman site in the south Huron Basin (Deller 1976; Fox et al. 2015). Plano forms like these are considered taxonomically to be late Paleoindian despite being their apparent contemporaneity with south early Archaic notched point forms (Fox et al. 2015). Dibb (2004) has interpreted the Madina phase Deavitt site as ideally situated on a treeless hilltop with a commanding view of the surrounding area. Caribou may have been intercepted as they water crossings between Deavitt and the adjacent uplands. The interpretation of Madina phase peoples selecting campsites for their advantageous positions with regards to caribou exploitation implies a similar lifeway to earlier
Paleoindians groups.
2.10.B Late Paleoindian Mobility and Raw Material Procurement
Madina phase lithic procurement may show evidence of settling in processes. Madina points manufactured on Trent, Collingwood and Onondaga cherts hint at a more generalized toolstone procurement pattern (Dibb 2004: 156). 14 different types of lithic raw material were recovered from late Paleoindian components in the Queensville-Keswick area (Dibb 2004: 131). The assemblage includes many lithic materials not utilized during the early Paleoindian period. The most common material noted is Balsam Lake chert. Balsam Lake chert outcrops at points along the shore of its namesake as well as appearing as nodules in nearby lakes (Eley and Von Bitter
1989). The use of more local cherts points towards a shift from the use of more massive, high quality cryptocrystalline raw materials seen in earlier times (Goodyear 1989; Haynes 1980: 118;
Meltzer 1984).
43
The Hi-Lo phase raw material acquisition patterns and mobility stand in contrast to early
Paleoindian times. Ellis and Deller (1982: 7) note that Hi-Lo materials are overwhelmingly manufactured on Haldimand or Kettle Point chert. This contrasts with the chert preferences of earlier groups and can be seen as a shift towards the use of more local materials as opposed to the primary use of distant cherts by earlier Paleoindian groups. Onondaga and Collingwood chert are considered to be of higher quality than Haldimand chert and thus the preference of Hi-Lo peoples for Haldimand chert cannot be explained by a desire for higher quality material (Ellis 2004a: 61).
The use of lower quality Haldimand chert is not consistent with early Paleoindian preferences for high quality cryptocrystalline materials (Goodyear 1989; Haynes 1980: 118; Meltzer 1984)
Haldimand chert and Kettle Point chert outcrop on opposite ends of southern Ontario
(Moerschfelder 1985: 8; Fox 2009: 362). The predominant usage of these cherts in place of northerly Collingwood chert suggests a change in mobility patterns. Gainey and Parkhill groups have been demonstrated to have typically followed a north-south movement pattern; the same cannot be said about Hi-Lo groups. North-south mobility patterns seem to have terminated along with the use of Collingwood chert. Hi-Lo groups must have had an east-west mobility orientation in order to enable the exploitation both Kettle Point and Haldimand chert. This interpretation assumes direct acquisition of toolstone as embedded within Hi-Lo movement.
44
Chapter 3
Cultural Transmission Theory and Cultural Change
3 Explaining Biface Variability
The intent of this thesis is to characterize and explain variability in Hi-Lo point size in Ontario and New York. Biface variability is a topic of considerable interest in archaeology. The appearance, florescence and distribution of a bifacial point form coupled with its subsequent decline and disappearance provides archaeologists with the opportunity to examine technological systems and test hypothesized causal factors driving cultural change.
Analyzing variability within existing point typologies provides an opportunity to test hypotheses concerning regional subsistence strategies and settlement patterns. For example, analyses of point form have been used to infer markedly different adaptation strategies for Clovis peoples. Regional variation in Clovis point form has been used to support the regional environmental adaptation hypothesis. This model proposes that Clovis groups altered the tool form in response to regional environmental conditions. Changes in prey caused variation in Clovis toolkits and point form across the continent (Buchanan et al. 2014: 146).
The continent-wide adaptation hypothesis argues that Clovis point form does not exhibit significant regional variation and is thus indicative of a widespread inter-regional adaptation
(Haynes 1964; Kelley and Todd 1988). A continent-wide Clovis adaptation forms a central part of the high technology forager model. The early Paleoindian toolkit is interpreted as flexible and capable of hunting differing prey across the diverse environments of North America. A toolkit with inter-environmental utility is thought to have aided colonization efforts and enabled the extreme annual range mobility of the early Paleoindians. This argument is expressed by Kelly and
45
Todd (1988: 236) who argue that “frequent range shifts may not have been conducive to in situ development of regional projectile point styles.”
The development of stylistic regionalization during the late Paleoindian period is thought to be a result of processes associated with settling in. Regional variation has been used to infer the development of social boundaries and the formalization of social interaction mechanisms in the late Paleoindian period (Anderson 1995; Koldehoff and Walthall 2004, 2009; Koldehoff and
Loebel 2009; Walthall and Koldehoff 1998).
The purpose of this chapter is to review theoretical and empirical work on the subject of biface variability, focusing on contributions that have informed this thesis’ approach to the interpretation of the spatial patterning of variability among lithic bifaces classified as Hi-Lo
3.1 Risk
Responses to risk of resource failure been considered as a factor driving technological variability
(Hayden et al. 1996; Torrence 1989; Collard et al. 2005; Nelson 1991; Bamforth and Bleed 1997).
The idea of risk as driving technological variability has its roots in Torrence’s (1983) study of the relationship between technology and time stress. Torrence proposed using latitude as a proxy for time stress with the assumption that “all other things being equal (e.g., altitude, rainfall) the length of the growing season for plants decreases on a global scale with increasing latitude” (Torrence
1983: 14). As latitude increases the number of edible plants will typically decrease, forcing hunter-gathers to depend on animal resources which are less predictable and demand greater temporal investment. Torrence hypothesized that as time stress increased, hunter-gathers would produce more specialized tools and increasingly complex toolkits.
Later Torrence (1989, 2000) moved away from the idea of time stress in favour of the notion of risk of resource failure. Risk is defined by Torrence (2000: 77) as “made up both of the probability of not meeting dietary requirements and the costs of such a failure.” Groups that
46 experience a high risk of resource failure can be expected to produce complex toolkits in order to minimize this risk. Conversely, those groups inhabiting plentiful, predictable environments where the risk of resource failure is low will produce less complex toolkits due to the minimal amounts of risk.
The concept of risk of resource failure has been central to the development of tool design theory.
Bamforth and Bleed (1997) view risk as a function of reliability and predictability while Keene
(1981) sees risk as a probability of failure or loss in a subsistence system. The idea of risk is seen as a mechanism of cultural change for archaeologists working within the framework of human behavioral ecology. Unpredictable resource availability or limited access to resources obligates people to make decisions to lessen risk and may precipitate the development of a novel technological strategy in order to mitigate is effects (Winterhalder et al. 1999).
Hi-Lo is thought to occur at a time of environmental change, 10,300 – 9,900 BP, during the
Pleistocene-Holocene transition (Goodyear 1982, Koldehoff and Loebel 2009: 138-139; Jennings
2010; White 2013; Ellis 2004a, Ellis et al. 2011: 536). The ongoing changes to landscape and floral and faunal communities associated with the shift to Holocene environments would have contributed to an increasingly uncertain environment where established risk mitigating strategies may no longer fit the needs of the population.
The decline of a spruce and subsequence florescence of pine and deciduous trees at this time may have produced an environment unsuited to the lifeways of presumed ancestral early Paleoindian peoples (Karrow and Warner 1990; McAndrews 1981, 1994; Yu 2000; Julig and Beaton 2015).
Changing environmental conditions may have induced the Hi-Lo knappers to develop a more utilitarian hafted bifacial tool to cope with the variable transitional environment.
47
3.2 Prey Availability
The risk hypothesis has its roots in the technological studies of Oswalt (1973, 1976) which have been influential for later studies of hunter-gatherer toolkit structure (e.g. Collard et al. 2005, 2011;
Shott 1986; Torrence 1983, 1989, 2000; Bamforth and Bleed 1997)
Oswalt’s work focused on the relationship between toolkit structure and composition and the nature of hunter-gather food acquisition. He developed a classificatory framework for subsistence related tools allowing for cross-cultural studies of toolkit structure. Toolkit comparison focused on tool forms that Oswalt identified as subsistants which were further classified as either instruments, weapons or facilities. Certain forms of Hi-Lo points could easily be classified as a weapon “a form that is handled when in use and is designed to kill or maim species capable of significant motion” (Oswalt 1976: 79) however, not all Hi-Lo forms neatly fit into the category of weapons. Toolkit structure was evaluated in terms of the total number of subsistants (STS), total number of technounits (TTS) and average number of technounits per tool (AVE). A ‘technounit’ is defined by Oswalt (1976: 38) as an “integrated, physically distinct, and unique structural configuration that contributes to the form of a finished artifact”
Oswalt (1976) postulated that toolkit structure was influenced by the nature of hunter-gather diets.
He argued that there was a relationship between a hunter-gatherer population’s degree of reliance on mobile resources (e.g. migratory caribou) and toolkit complexity. Those populations which draw primarily on immobile terrestrial resources like plants should be expected to employ a less complex toolkit. The mobility of a terrestrial resource obligates hunter-gathers to create more complex toolkits to mitigate the effects of unpredictable access to mobile prey. Oswalt’s dietary
(1973, 1976) hypothesis and Torrence’s (1989, 2000) risk hypothesis are based upon the idea that the availability of a desired resource impacts the design of tools.
48
The work of Newby et al. (2005) has demonstrated that in the Northeast/Maritimes region, the use of lanceolate fluted points coincides with environmental conditions capable of supporting large herding mammals. The shift towards mixed, deciduous forests and the increased floral biomass associated with the warmer and wetter southern Ontario Holocene environment (see Julig and
Beaton 2015: Figure 5.1 for a current reconstruction circa 9000 BP) would have been more suited to non-herding solitary cervids, such as woodland caribou, moose or deer (Ellis and Deller 1990).
In absence of predictable prey (i.e. migratory caribou following a known route) and food storage technology, large population aggregations as inferred for the Parkhill site (see Deller and Ellis
1992b; Storck 1997) would generally not have been possible.
In contrast with earlier fluted forms, Hi-Lo points are thought to have been commonly employed for a non-projectile use, functioning as drills, side-scrapers or knives in a manner which frequently impacted blade shape (Ellis and Deller 1982, 2014; White 2012). Ellis (2004a) has suggested that the incipience of notched Hi-Lo may represent a shift to a more multi-purpose tool.
The functional significance of these changes in Hi-Lo morphology and their relation to changing floral and faunal communities are not yet fully understood. Hi-Lo tools may have been designed and utilized to meet the needs of a population practicing a subsistence strategy more directly conditioned by local floral communities than that of early Paleoindians. A more generalist subsistence strategy drawing on mobile, solitary cervids (e.g. deer or moose) as well as immobile, local floral resources would be difficult to evaluate using Oswalt’s prey hypothesis. Without the benefit of detailed floral or faunal remains associated with Hi-Lo occupations it is difficult to evaluate their relative importance to subsistence and assess Hi-Lo variability in relation to the expectations of the prey availability hypothesis.
49
3.3 Mobility and Raw Material Management
The idea that hunter-gatherer mobility influences toolkit structure and tool design was first proposed by Shott (1986). Shott saw carrying costs associated with tool transport as influencing the design and number of tools a population will employ. Shott’s model predicts that groups with a high degree of residential mobility will favour tools that are light, compact, easily transportable as well as long-lasting and renewable. Highly mobile groups are also more likely to employ less complex toolkits than less mobile groups. His model suggested a negative correlation between the number of residential moves per annum and toolkit diversity.
To evaluate the model Shott (1986) preformed two sets of analyses. The first analyzed the toolkits of 14 historically documented hunter-gatherer populations working within the comparative framework established by Oswalt (1973, 1976). Parametric and nonparametric analyses were used to assess the relationship between the number of subsistants and average number of technounits per subsistant with measures of residential mobility. The term technounit is used to denote each
‘integrated, physically distinct, and unique structural configuration that contributes to the form of a finished artefact’ (Oswalt 1976: 38). Mobility variables included the number of residential moves per year, annual distance traveled, average length of residential movement and the total area occupied in square kilometers.
Results from the first set of analyses were mixed. Toolkit diversity and residential move frequency per annum showed a significant negative correlation. This result supports the notion that populations with greater frequencies of residential moves are more likely to employ a lesser number of subsistants than peoples of lesser mobility or sedentary populations. The analysis of toolkit diversity showed no significant correlation with distance covered per annum and toolkit complexity, nor did toolkit complexity show a definite relationship to either frequency of residential moves or average distance per residential move.
50
Shott’s (1986) second set of analyses considered relationships between the technological variables and measures of logistical mobility. Logistical mobility variables included the number of days spent in the primary winter camp and intensity of land use. This second set of analyses also yielded mixed results. The analyses revealed a positive correlation between toolkit diversity and the number of days spent in the primary winter encampment. Toolkit diversity showed no significant correlation with the intensity of land use, nor did toolkit complexity correlate with the number of days spent in the primary winter encampment or intensity of land use. The relationship between the technological variables and Shott’s two environmental parameters, effective temperature and net primary productivity, were not found to be significant.
Shott’s model predicts that groups with a high degree of residential mobility, as is assumed for
Hi-Lo groups, will favor tools that are light, compact, easy to transport, which are long-lasting and renewable. Some support for this hypothesis can be found in previous studies of Hi-Lo morphology. Hi-Lo points are consistently resharpened, presumably as a functional modification related to their status as multi-functional tools (Ellis 2004a; Ellis and Deller 1982, 2014; White
2012). Evidence of resharpening is very common on Hi-Lo points suggesting that these tools were both long lasting and renewable. Ellis and Deller (2014) have recently examined Hi-Lo reduction and resharpening using a life history approach. This approach emphasizes the plasticity of Hi-Lo form over the course of its use life and affirms their status as long lasting, renewable tools.
The fact that most Hi-Lo points are isolated surface finds suggests some element of mobility likely related to targeted resource acquisition, whether floral or faunal. While likely less mobile than early Paleoindians, Hi-Lo groups were still exploiting large territories (Ellis 2011).
Interpreting Hi-Lo morphology in terms of mobility estimates and carrying costs associated with tool design offers an important perspective for a time period associated with environmental change and shifting raw material preferences.
51
3.4 Cultural Transmission Theory
Cultural transmission theory provides a framework for explaining artifact variability, similarity, and relatedness. Cultural transmission theory is based upon the notion that similarities in behavior and artifact form may be the result of information exchange through non-genetic mechanisms
(Eerkens and Lipo 2007: 240). Cultural transmission is an evolutionary process because the inherent nature of cultural transmission precludes perfect duplication thus ensuring change
(Eerkens and Lipo 2005; Bettinger 2008). The development of evolutionary approaches to archaeology has led to an intense interest in cultural transmission among some evolutionary archaeologists (See chapters in O’Brien 2008; Collard et al. 2008; Eerkens and Lipo 2005, 2007;
Hamilton and Buchanan 2009; Bentley and Shennan 2003; Mesoudi 2011; Mesoudi and O’Brien
2008; O’Brien et al. 2014; Richardson et al. 2009).
However, cultural transmission should be regarded as a unique process from biological genetic transmission and caution should be used in drawing strict analogies between the two. An important difference between genetic and cultural transmission is the fact that cultural evolution may occur much more rapidly than genetic evolution. Cultural evolution differs from genetic evolution as includes the possibility that individuals actively seek out new, more advantageous cultural variants in response to changing circumstances (Richardson et al. 2009).
3.4.A Modes of Transmission
Transmission in hunter-gatherer societies can be heuristically viewed from the perspectives of vertical, horizontal and oblique modes of cultural transmission (Mesoudi and O’Brien 2008;
O’Brien et al. 2012, 2013; Jordan and Shennan 2003). Cavalli-Sforza and Feldman (1981) employed the term ‘vertical’ to describe cultural transmission between biological parents and their offspring. The transmission of cultural traits between members of the same biological generations
52 may be referred to as being ‘horizontal’. While social learning between individuals of different generations is referred to as ‘oblique’ (O’Brien et al. 2012; Eerkens and Lipo 2007).
3.4.B Guided Variation and Common Descent
Common descent in cultural transmission theory refers to the notion that similar cultural manifestations may owe their similarity to a mutual antecedent. Because cultural information is passed between individuals, similarity in artifact form may sometimes be attributable to a common source of cultural information (Eerkens and Lipo 2007: 241). Variability in established tool designs can be tied to an individual’s experimentation with learned tool designs. Innovations of this nature are referred to as resulting from guided variation (Boyd and Richerson 1985: 82;
Richerson and Boyd 2004: 69).
3.4.C Reconstructing Heritable Continuity through Archaeological Phylogenetic
The early 21st century has seen a growing appreciation for biological method and theory as an aid to archaeological explanation (Barton and Clark 1997; O’Brien and Lyman 2002; O’Brien 2008,
Buchanan and Collard 2010). To understand relationships of heritable continuity between bifacial point forms developed through guided transmission, archaeologists are increasingly employing cladistics as a means to reconstruct phylogenetic relationships between archaeological taxa (e.g.
Buchanan and Collard 2007; Eerkens et al. 2005; O’Brien et al. 2014). Phylogenetic studies are rooted in Darwin's (1859) theory of descent with modification. While initially conceived as a means to explain species diversification, the concept of descent with modification may be applied to changes in artifact form (O’Brien et al. 2001). Descent with modification allows archaeologists to see that artifacts are related, not because they are similar, but similar as a result of a possible relation through common descent. Archaeological phylogenetics seek to emphasize heritable continuity in cultural change and the continuous nature of change. Heritable continuity can be conceived as tool design variants which arise through guided variation (Boyd and Richerson
53
1985: 82; Richerson and Boyd 2004: 69). In keeping with anthropological archaeology’s goal of explaining human cultural diversity, phylogenetic methods seek to shift emphasis from historical continuity – A before B – to heritable continuity — A precedes and gives rise to B (O’Brien et al.
2014: 116). Evolutionary archaeologists are increasingly operationalizing cultural transmission theory through the employment of cladistics as a means of assessing phylogenetic relationships between projectile point forms.
3.5 Drift and Cultural Change
The concept of drift is important to evolutionary approaches to biological and cultural variation.
Drift and selection are often considered to be the fundamental sources of cultural change in evolutionary archaeology. Although drift is commonly cited as a mechanism of cultural change in evolutionary archaeology, it remains an inconsistently defined term (Collard et al. 2008).
The interpretive implications of differential definitions of drift may be seen through comparison of definitions put forth by O’Brien and Lyman’s (2000) and Wilhelmsen (2001). Drift is defined by O’Brien and Lyman (2000: 399) as “random changes in trait frequency in a population resulting from the vagaries of transmission.” Wilhelmsen (2001: 100) views drift as the differential reproduction of organisms as a result of “sampling and error and the stochastic patterns of transmission.” Sampling effects during biological reproduction, which will be related to group fitness, will cause some phenotypes to increase in frequency and others to decrease due to chance (Bettinger et al. 1996). Wilhelmsen’s definition focuses on the differential reproduction of organisms while O’Brien and Lyman (2000) omits any mention of organisms. Instead, they focus on artifact form rather than the reproduction of flesh and blood organisms. This focus allows O’Brien and Lyman to avoid making specific the mechanism of cultural change and allow for the possibility that genetic and cultural evolution involve different mechanisms (Collard et al.
2008).
54
3.5.A Copying Error as a Source of Drift
Applying the concept of drift to cultural transmission theory, Hamilton and Buchanan (2009, 280) have advanced a definition of “drift as a measurable change in point form because of neutral stochastic processes caused by sampling effects that occur as the result of cultural transmission in finite, naturally fluctuating populations.” The transmission of cultural information for lithic tool design between individuals is not a perfect process. During the transmission process various factors influence the nature of the copying to resulting in imperfect duplication. The biomechanical and cognitive limitations of humans to assess and faithfully replicate the physical form of material culture ensures the possibility of unintentional design modifications each time an artifact is created (Eerkens and Lipo 2005; Hamilton and Buchanan 2009; Neiman 1995; Eerkens et al. 2014). The term ‘copying error’ is used to refer to the inability of humans to replicate objects of material culture (White 2013a). Copying error forms an important element of cultural transmission, and the accumulated effects of copying error have been found to cause drift in culturally transmitted technologies such as bifacial lithic tools (Eerkens and Lipo 2005; Hamilton and Buchanan 2009).
3.5.B Innovation Causing Drift
Technological innovation and experimentation is a critical component of human cultural evolution (Richardson et al. 2009). While design variations may be intentional they can also be the result of accidental discovery (Eerkens and Lipo 2005: 319).Though sometimes used synonymously, invention and innovation may also be used to describe different concepts
(Fitzhugh 2000: 128). Invention, following Fitzhugh (2000: 128) is “the development of a novel idea with its attendant material, practical, and informational components.” Inventions are considered as novel and untested ideas, tools and/or methods. Innovation (Fitzhugh 2000: 128) refers to “the process of testing and putting into practice an invented method/device.” Similarly, an innovation is an invention that has been put into use and is no longer novel and untested. An
55 important distinction between these two closely linked concepts is the difference between the incipience of novelty and the processes involved with its subsequent implementation and adoption.
O’Brien et al. (2012: 551) see innovation as “a source of novelties similar to genetic mutation and recombination.” O’Brien sees innovation as both the act of invention and the implementation of a novelty. Archaeologists working within a drift-selection framework often see innovation as random and undirected by selection (Dunnell 1978; Neiman 1995; O’Brien 2012: 551). While innovation may be random, the subsequent adoption and persistence of an innovation is subject to selective pressures and the processes of drift associated with the transmission of culture (Bettinger and Richerson 1996).
3.5.C Drift and Effective Population Size
The effects of drift are also closely tied to overall population characteristics. Since drift is a result of sampling, its effects are increased in small populations with fewer cultural emitters (see
Henrich 2004; Shennan 2001). The increased effects of drift in small, isolated groups is known as the founder effect. These populations may not accurately reflect the variation found in the larger parent population. They may include only a subset of this total cultural variation found in the larger population. The founder effect may be expressed as the increased rate of drift in small populations combined with the differential representation of cultural variants retained in a breakaway population (Buchanan and Hamilton 2009).
Total effective population size has been hypothesized to influence the structure of hunter-gatherer tool technology. The relationship between population size and cultural innovation has been explored by Shennan (2001) who modelled the rates of innovation and adoption of beneficial or deleterious traits in variable populations. Shennan’s work is based in an evolutionary approach known as dual inheritance theory (Bettinger 1991; Boyd and Richerson 2005; Cavalli-Sforza and
56
Feldman 198; Shennan 2000, 2001, 2002). Duel inheritance theory sees cultural evolution as sharing many of the same properties as genetic evolution but acknowledges some differences.
Cultural evolution is seen as occurring at a faster rate than genetic evolution. This increased rate of change is due the fact that cultural information is often transmitted on a much shorter scale as individuals actively seek out desirable cultural variants to improve design many times over the course of their lives.
Shennan (2000) used two models for simulations modified from population genetics models initially formulated by Peck et al. (1997). In Shennan’s models, mutations (innovations) were coded as either beneficial or deleterious. Each mutation resulted in a small change in the cumulative fitness value of the individual. Shennan’s first model restricted transmission from parent to offspring (vertical transmission) while the second model allowed for transmission between an individual and unrelated individuals of the younger generation (oblique transmission).
Shennan (2000) found a strong correlation between the mean fitness of the simulated population and effective population size. The first model indicated a 10, 000 times increase in the mean fitness value of the population as effective population size increased from five to 50. In a simulation of the second model where cultural innovations were obliquely transmitted five percent of the time, population mean fitness value increased 1, 000 times as the effective population size increased from five to 25. These results indicate that populations become less subject to the sampling effects causing drift as population size increases. When populations are larger there is a greater chance of fitness-enhancing innovation being retained by the population.
Large populations are more likely to maintain advantageous innovations while small populations are more prone to losing fitness-enhancing innovations due to drift. Shennan’s (2000) findings suggest that small populations are likely to employ less complex toolkits while large populations may be expected to employ larger and more complex toolkits.
57
3.6 Hypotheses
This study aims to interpret Hi-Lo biface variability from an evolutionary perspective using cultural transmission theory. The following section lays out hypotheses for the interpretation of
Hi-Lo variability within the evolutionary drift-selection framework and further develops sub- hypotheses informed by cultural transmission theory relating to the geographic scale of Hi-Lo social learning. Sub-hypotheses relate to spatial patterning effects predicted for different scales and intensities of regional interaction during the late Paleoindian period. These sub-hypotheses use cultural transmission theory to test for archaeological manifestations of processes central to the regionalization hypothesis; decreased mobility and the development of territoriality and other social barriers to inter-group interaction.
3.7 Selection (Risk/Function)
Following O’Brien and Lyman (2000: 404), selection is defined as a “process by which certain forms in a population that are better adapted to a particular environment increase in proportion to less well-adapted forms.” Selective pressure often favours functional technological variants which confer an adaptive advantage to their users, act to increase individual and group fitness and reduce the risk of resource failure. This means that the proliferation of a functional design variant is most likely to occur within areas where that technological variant provides a selective advantage.
While this research uses an evolutionary approach, it does not employ the style-function dichotomy of Dunnell (1978). I conceive of style as having functional aspects in a manner similar to Bettinger, Boyd and Richardson (1996). This means that stylistic expression can have indirectly functional aspects. For example, indirectly functional aspects could take the form of group emblemic style which acts as a type of signaling. While group emblemic style may display geographic patterning, it is unlikely that an individual’s assertive style could be discernable among Hi-Lo points without a highly detailed study of Hi-Lo points known to be
58 contemporaneous (Wiessner 1983). Consistent with Bettinger, Boyd and Richardson (1996) stylistic evolution is considered to be governed by cultural evolutionary processes which may result in non-random patterning of stylistic variation. Therefore, the spatial patterning of stylistic variants may in some cases be indistinguishable from that of variants whose patterning is conditioned primarily by processes of selection.
Decreasing residential mobility would mean that groups’ subsistence strategies would be more directly conditioned by their immediate environments. If decreasing residential mobility forced groups to adapt their subsistence strategies to pressures imposed by local environments, selection will primarily drive the variation in the Hi-Lo technological system. However, selection is acknowledged to be difficult to identify and distinguish from other processes that generate technological variants (Bettinger et al. 1996). If there is evidence of regional variation in Hi-Lo bifaces, this might be attributable to selection if (1) it can be demonstrated that the variation relates to differences in biogeographic factors related to subsistence practices; (2) the variation can be attributed to design optimization related to regional differences in raw material. Future developments in evolutionary archaeology will likely lead to more refined methods for identifying selection in the archaeological record. For this reason, the above criteria for a positive identification of selective pressure in Hi-Lo morphology should not be considered exhaustive.
3.7.A Innovation as a Response to Risk
While selection may be difficult to identify in the archaeological record, risk defined by Torrence
(2000: 77) as “the probability of not meeting dietary requirements and the costs of such a failure” may provide a framework for interpreting the role of selective forces in the generation of variability in bifacial tools.
Boyd and Richerson (1985:116) have proposed that predicable environments discourage the incidence of innovation. In predicable environments it holds that established tool designs have
59 likely been selected as successful variants. However, in unpredictable environments personal experimentation will be more frequent as individuals attempt to adapt tool design to mitigate risk.
Unpredictable environments encourage individual innovation at the expense of social learning through vertical or oblique transmission. The risk mitigating strategies previously employed by an elder generation may no longer satisfy the needs of the younger generation.
Empirical support for the risk hypothesis is provided by the work of Collard and colleagues
(2005, 2011, 2012, 2013). Collard et al. (2005) applied step-wise multiple regression analysis to technological and ecological data for 20 hunter-gatherer populations to compare factors which have been hypothesized to influence hunter-gatherer toolkit diversity and complexity. The hypotheses tested included 1) the nature of the food resources; 2) risk of resource failure; 3) residential mobility; and 4) population size. The only hypothesis which was supported by the analysis was Torrence’s (1989, 2000) risk hypothesis. The only predictor variables significantly influencing toolkit structure were LET (natural logarithm of effective temperature) and LNAGP
(natural logarithm of net above ground productivity). Both of these predictors were included considered as proxies for risk. Of the four hypotheses tested by Collard et al. (2005, 2011, 2012,
2013) the only one to enjoy empirical support is Torrence’s (1989, 2000) hypothesis that risk drives hunter-gatherer decision-making about the toolkit diversity and complexity.
Without a highly refined paleo-environmental reconstruction local responses to risk may be unrecognizable in the Hi-Lo record. Hi-Lo forms are thought to have been in use for several hundred years at a time of major environmental change brought on by the onset of the Holocene.
Any attempt to explain variability in relation to contemporary local environmental conditions will likely be confounded by the lack of sufficiently fine-grained paleo-environmental data and temporal refinement for changes in Hi-Lo design. In the future, combining the data presented in this study with Hi-Lo data presented by White (2013: Figure 5) for the midcontinent United States may provide continuous Hi-Lo data for a suitably large geographic area which could be used to
60 broadly test for technological responses to risk. This effort could be done using latitude as a proxy for risk (Torrence 1983: 14). At present it may be premature to test for risk and other selective pressures in the Hi-Lo record.
3.8 Drift: Accumulated Copying Error and Innovation in an Internally Unbounded Learning System
Variability in Hi-Lo technology may be driven by the accumulated effects of copying error over time. Since Hi-Lo technology is thought to persist for a number of centuries (See Ellis 2004a) a significant amount of transmission error may be expected to accumulate as a result of the
“vagaries of transmission” (O’Brien and Lyman 2001: 399).
Variation arising from copying error is a mode of cultural change considered as evolutionary drift.
The ability of an individual to faithfully imitate tool design is hampered by biomechanical and cognitive differences inherent to the human condition. Due to the imperfect nature of cultural transmission, the effects of drift will be present in all circumstances lacking a strong conformist tradition and the ability to perfectly replicate a design template. Over time, accumulated effects of copying error can alter the initial design template used for tool production such that later individuals aim to create tools which are significantly different than those of their predecessors.
Thus, the imperfect copying intrinsic to cultural transmission is considered to be the default source of cultural change in the absence of selective force.
Social learning in hunter-gatherer societies most frequently takes place within one’s immediate group of cohabitants. Since the effects of drift driving cultural change are amplified in smaller populations, a group of residentially mobile, cohabitating Hi-Lo knappers can be assumed to have had a high rate of exposure to novel designs variants when they interacted with external groups of
Hi-Lo knappers. Inter group social learning provides knappers with opportunities to copy design variants which were developed outside of their own immediate social group. The conspicuous use of a common place of return (i.e. primary chert outcrops) could be an indicator of an area where
61 inter-group cultural transmission frequently occurred. Continual sharing of knowledge and design variants between groups would have a homogenizing effect on tool design between knappers creating Hi-Lo points.
Fragmentation of the effective population of knappers through the development of social barriers as inhibitors to cultural transmission is predicted by cultural transmission theory to have a recognizable effect on Hi-Lo tool design. Population fragmentation is predicted to encourage heterogeneity in Hi-Lo design. Conversely, frequent interaction and transmission of ideas and designs between groups of knappers manufacturing Hi-Lo points is predicted to have a homogenizing effect on the Hi-Lo design template, as design variants are frequently shared within the larger population of Hi-Lo knappers.
In absence of selective pressure or social/geographic barriers to cultural transmission, design variation is assumed to be the result of the accumulated design deviations arising from copying error compounded over time. Consequently, this hypothesis predicts that if there is no regional patterning in Hi-Lo biface variability, then variability can be interpreted as resulting from accumulated random copy error in a learning system unconstrained by territoriality and social/geographic boundaries.
3.8.A Inherent Synchronous Design Variation
Variability may be inherent in the design template of Hi-Lo. What constitutes a Hi-Lo point in the minds of the maker may not be rigorously defined as different persons may view Hi-Lo design in ways only loosely similar to each other. If the Hi-Lo template was not rigorously defined, artifact morphology may display significant morphological variability while still being considered to fit within the design template of Hi-Lo.
This view sees Hi-Lo as a non-conformist technological system with few constraints on transmission error. If variability is inherent in Hi-Lo design then a considerable range of design
62 variability may characterize the forms produced by an individual over even a limited amount of time. If variability is inherent in the Hi-Lo design template it is expected that variability within the Hi-Lo learning system will be spatially random. Patterning predicted to arise from synchronous design variation is difficult to distinguish from the random patterning of variability predicted by a scenario which sees Hi-Lo design variation as driven by accumulated copying error in an internally unbounded learning system. While both scenarios predict random spatial patterning for Hi-Lo variability they are distinguished by the role played by time. Synchronous design variation implies contemporary knappers producing widely varying forms while the latter sees time as the variable which accounts for increasing design deviation.
3.8.B Geographic Regionalization of Variation in a Bounded Learning System
The mechanisms of drift may cause differentiation in Hi-Lo design between sub-populations separated in space. Regional differences in artifact design may arise though drift in the absence of selective pressure. A non-random spatial signature for biface variability is likely attributable to the effects of drift unless it can be demonstrated that variability maps on to economic parameters which have exerted selective pressure.
Cultural transmission theory predicts that hunter-gatherers will employ a narrower range of material culture than agricultural societies, due to the limited effective population size imposed by their lifeway (Eerkens et al. 2014). Elsewhere, growth in effective population size with the onset of the Holocene has been suggested as the catalyst for the increased diversity of material culture during the Mesolithic and Neolithic (Shennan 2001). Supporting research by Henrich (2004) linked the loss of a number of complex tool technologies from the Tasmanian tool repertoire to a decrease in effective population size when rising Holocene sea levels cut Tasmania off from the
Australian continent. The work of Shennan (2000, 2001) suggests that smaller populations are more subject to the sampling effects causing drift than larger ones. Effective population size
63 should be considered both in terms of sheer population and social organization which govern the scale and intensity of regional interaction patterns.
An individual’s opportunities to copy different designs is limited by the rate by which that individual is exposed to differing designs. Opportunities for cultural copying are limited by social network size (White 2013b). In hunter-gather societies, the greatest amount of interaction occurs within an individual’ immediate group; often learning from those in their immediate family. A large proportion of technological knowledge is acquired through vertical transmission (parent to child) or oblique transmission (elder-younger). Thus cultural transmission can be assumed to occur most frequently at a small geographic scale.
While most cultural transmission occurs within an individual’s immediate group, interactions with other groups present important opportunities to copy new designs. The effects of neutral drift upon Hi-Lo design would have been exacerbated by the wide geographic distribution of Hi-Lo making peoples. This result is because individuals are assumed to have less frequent opportunities for interaction with distant groups compared to those within their own group (Eerkens et al.
2014). This assumption is supported by social network modelling work by White (2013b) suggesting that instances of inter-group encounters are inversely correlated with social network distance. The development of social or geographic impediments to cultural transmission would act to segment the effective population of the social learning system represented archaeologically by Hi-Lo points. Corresponding decreases in residential mobility and inter-group interaction predicted by the regionalization model increase the sampling effects of drift in separated Hi-Lo groups and intensify inter-group design variability.
Fragmentation of the effective population would result in founder effects intensifying inter-group variability (Eerkens 2000; Buchannan and Hamilton 2009). The term founder effect is used to express increased rate of drift in small populations combined with the differential representation
64 of cultural variants retained in breakaway populations (Buchanan and Hamilton 2009; Henrich
2004; Shennan 2000, 2001).
The development of social barriers characterized by a lack of interaction is predicted to manifest archaeologically as a heterogeneous range of Hi-Lo design variability. Limitations on the geographic scale of social learning should be represented in the archaeological record by variability in Hi-Lo point size and configuration which displays significant spatial autocorrelation.
If processes relating to regionalization and settling in are effecting the scale of social learning during the late-Paleoindian period Hi-Lo point size should display non-random spatial patterning.
3.8 Design Variation Driven by Raw Material Constraints
The execution of a mental Hi-Lo design template may be influenced by raw material constraints.
Differences between qualities of raw material may influence the ability of the makers to produce desired tool forms (Bamforth 1991; Long 2004; Grimm 2000). Since Hi-Lo points have been found to be made from a variety of toolstones (see Ellis and Deller 1982) raw material differences should be considered as a potential source of Hi-Lo variability. Design considerations arising from raw material constraints may confound Hi-Lo knappers’ attempts to reproduce a desired design template. If differential raw material constraints are significant enough to constrain Hi-Lo design then variability in Hi-Lo design should be correlated with raw material.
65
Chapter 4
Methods and Sample Composition
4 Quantitative Assessment of Shape and Size
The study of variation in material culture is among the most enduring archaeological endeavors.
The overall form of an object may be described in terms of size and shape (Lycett 2009: 82).
Shape refers to the geometric properties of an object that are independent of the object’s overall size (Mitteroecker and Gunz 2009: 237).
Lithic technology has long been the centerpiece of Paleoindian studies. Given the rarity of associated organic remains from the time period, Paleoindian researchers have often been required to rely on the sequencing of bifacial point “types” to act as cultural-temporal markers
(e.g. Deller and Ellis 1988; Justice 1987; Anderson 1990; White 2013). Clearly defining and understanding the variability between and within Paleoindian point types continues today as an active area of research aided by the development of advanced digital analysis methods (e.g.
Thulman 2012; Morrow and Morrow 2009; Sholts et al. 2012; Shott 2014; Buchanan and Collard
2007. 2010; O’Brien et al. 2014; White 2013).
The importance of point typology to Paleoindian archaeology is paramount. Analyses of point types has been used by researchers to make inferences and construct models that today form the basis of our knowledge of the Paleoindian period. The importance of point types to Paleoindian archaeology has recently been underscored by Gingerich et al. (2014: 102) who notes the use of early Paleoindian fluted point types to:
1) Establish the Pleistocene antiquity of man in the New World (e.g. Figgins 1927, 1933).
2) Discuss late Pleistocene mobility and subsistence strategies (Frison 1989; Kelly and Todd
1988; Meltzer 1988: 4).
66
3) Infer population and colonization patterns (e.g. Anderson and Faught 1998; Anderson and
Giliam 2000; Goebel et al. 2008; Hamilton and Buchanan 2007; Mason 1962; Morrow and
Morrow 1999; Stanford and Bradley 2012).
4) Propose chronological and regional boundaries (Collard et al. 2010; Wright and Roosa 1966;
Goodyear 1982, 2006; Haynes et al. 1992; O’Brien, Darwent, and Lyman 2001)
5) Define distinct technological attributes that may relate to cultural or regional traditions (e.g.
Ahler and Geib 2000; Bradley et al. 2010; Ellis and Deller 1982, 2014; Gillespie 2007).
4.1 Multivariate Morphometrics
Morphometric studies express variability between forms in quantitative terms. Morphometrical analyses were first pioneered by biometricians but have recently seen application in archaeology
(i.e. Hamilton and Buchanan 2009; Buchanan and Collard 2010; Cardillo 2010; Gingerich et al.
2014; Thulman 2012). The term “morphometrics” was first coined by Blackith (1957) while studying the characteristics of locust swarming events. Blackith applied multivariate statistical methods to the basic carapace morphology of locusts to understand the morphological changes which signaled population explosions indicative of a swarming event. Today, morphometrics may be generally defined as an “interwoven set of largely statistical procedures for analyzing variability in size and shape” (Reyment 2010: 9).
Blackith’s approach to morphometrics, common in the 1960’s and 70’s is now referred to as multivariate morphometrics (Blackith & Reyment: 1971) or traditional morphometrics (Marcus,
1990; Reyment 1991). Morphometric studies at this time utilized linear distance measurements as well as counts, ratios and angles. Using these measurements, covariation in morphological measurements could be quantified and patterns of variation between specimens could be determined. Statistical analyses used in multivariate morphometrics included principal
67 components analysis (PCA), factor analysis, canonical variants analysis (CVA), and discriminant function analysis (Adams et al. 2004: 6).
Initial multivariate morphometrics studies were hampered by methodological limitations concerning the ability of researchers to capture all aspects of shape (Marcus 1990; Slice 2005;
Bookstein et al. 1985). Early morphometric studies often measured metric variables (e.g., maximum width, length) at locations not defined by homologous points which were unable to accurately capture shape (Adams et al. 2004: 6). Homology refers to “points of morphological correspondence (landmarks), which may be identified according to explicit and clearly defined rules.” (Lycett 2009: 81). The importance of homologous landmark placement is apparent through the shortcomings of traditional linear measurements to accurately capture shape. Adams et al.
(2004: 6) cite the example of the linear similarities between a tear drop shape and an oval to show the limitations of linear distance measurements (Figure 4). Maximum width and height measured on both shapes may be identical which would mask variation between the subjects.
Figure 4. Comparison of maximum length and width between an oval and teardrop shape
4.2 Geometric Morphometrics
Traditionally, morphometrics had been the application of multivariate statistical analyses to sets of quantitative variables such as length, width, and height. In the late 1980s and early 90s a shift occurred in the way morphometricians quantified physical structures of interest and how data was expressed.
68
The “new” approach heralded by Rohlf and Marcus (1993) is characterized by data recorded with the intent of capturing the geometry of the subject. Emphasis is placed on methods which capture the geometric relationships between morphological structures of interest and carrying this information forward through the analyses (Adams et al. 2004: 5). The emphasis on geometry led to the coining of the term ‘geometric morphometrics’ (Corti 1993) and the suggestions that this methodological shift represented a “revolution in morphometrics” (Rohlf & Marcus 1993)
Geometric data is recorded by the use of two dimensional (2D) or three dimensional (3D) landmarks applied to the study object (Rohlf and Marcus 1993: 129). Landmarks record coordinate data for a chosen point on an object. Coordinate data allows for the complete retention of shape geometry. Coordinate based approaches to landmarks are able to encode all the information detailing the distances and angles between landmarks in a way that traditional landmarks could not (Slice 2007).
Geometrical relationships between the landmarks are not inherent in their placement. The geometrical relationships between the landmarks can be understood through the application of the appropriate functions, whether in 2D or 3D. Thin-plate spline functions (TPS) are used to discern differences in landmark position on an object (Bookstein 1989, 1991). TPS functions are able to express the differences in the configurations of separate sets of landmarks as a continuous deformation. Measured parameters are then used as variables in univariate or multivariate statistical analyses (Rohlf and Marcus 1993).
4.2.A Landmark Classification
Landmarks may be classified as either anatomical, mathematical or semi landmarks. Anatomical landmarks mark points that correspond between subjects in a way that is both descriptive and explanatorily meaningful (Barceló 2010). Anatomical landmarks are considered to be homologous as they mark points of morphological correspondence. The homologous placement of
69 landmarks allow change in shape or attribute to be recorded relative to the changes in other structures within a subject (Lycett 2009). With regards to hafted biface analysis, anatomical landmarks may include the point’s tip, points marking the maximum width, constriction, etc.
Mathematical landmarks are points whose location on an object are determined by a mathematical or geometric property of the object. These landmarks are not considered homologous as they do not correspond to anatomical locations on the subject. Barceló (2010: 116) defines mathematical landmarks as “any set of points that are characterizable and searchable upon a surface.”
An alternate approach to landmark placement utilizes what are called pseudo or semi landmarks placed along the object’s outline or in positions between anatomical or mathematical landmarks to capture size and shape (Buchanan and Collard 2010; Barceló 2010). Semi landmarks are meant to sample a subject uniformly. Assessments of variability between objects measured with semi landmark data requires the presence of at least two mathematical or anatomical landmarks as reference points (Cardillo 2006).
4.3 Collection and Analysis of Metric Data
This section defines the methods used to describe patterns of metric variability among Hi-Lo projectile points. The projectile point sample (n=302) analyzed in this study was drawn from numerous academic, museum and private collections; as well as materials recovered through cultural resource management (CRM) investigations within Ontario. These Hi-Lo points were recovered through the work of innumerable academic, avocational and consulting archaeologists without whom this work would not be possible. The reader is directed to the acknowledgements section for a list of persons or institutions who made available collections for study.
Sample documentation was facilitated by the Ontario Ministry of Tourism, Culture and Sport
(MTCS) which made available a site register detailing the present locations of Hi-Lo collections in the province known to the MTCS. When possible, the author visited these collections in person
70
(Figure 5) in order to obtain scaled photographs of the points and collect additional information which may not have been included in the initial publications (e.g. weight, thickness or raw material). If collections were not available for personal inspection, scaled images and point data
Figure 5. Hi-Lo collections accessed were drawn from published research papers or technical reports. Other than weight and thickness, all measurements were drawn from 2D scaled images.
4.3.A 2015 Hi-Lo Sample and the Expanded 2015 Hi-Lo Sample
The assessment of bifacial form and raw material identifications was undertaken on a sample comprised of those points previously examined by Ellis and Deller (1982) as well as some of the points reported by Roberts (1985) and Tinkler and Pengelly (2004). Since not all points described in the literature are depicted in publications or available for in-person examination it is necessary to speak of a current 2015 Hi-Lo Sample, for which bifacial form was assessed, and an Expanded
2015 Hi-Lo Sample which includes raw material identifications for known points whose form was not able to be assessed as part of the current study. The Expanded 2015 Sample includes all points included in the 2015 Hi-Lo Sample as well as some additional points attested in the literature of
Ontario archaeology which were tied to specific locations or a known region within the study area
71 and had been previously assigned to a raw material type. The additional points included in the
Expanded 2015 Sample was necessary to provide the most accurate representation of Hi-Lo raw material usage patterns possible. This section provides a description of additional points included in the Expanded 2015 Hi-Lo Sample.
Roberts (1985: 83) reports a total of 24 Hi-Lo points from 18 unique location from the
Burlington-Oakville and Durham regions. These points were identified as being split 50% between Onondaga and Ancaster chert (1985: 86). Roberts’ regional study areas are depicted along with summary information concerning Hi-Lo finds as well as revised raw material identifications (see Ellis 2004a: 61) in Figure 6.
Figure 6. Study areas and Hi-Lo raw material Identifications in Roberts 1985
Ellis (2004a: 61) disputes Roberts’ (1985: 86, 211-212) raw material identification of Ancaster for the ENL 502 Hi-Lo site (1985: Fig. AIV-2) discovered during Roberts’ (1978, 1978, 1985) survey work along the north shore of Lake Ontario. Roberts identified Ancaster chert as the source material for the ENL 502 site’s assemblage. Ellis believes that this is due to Roberts being
72 unaware of the existence of Haldimand chert as Roberts’ work was published before relevant developments in Ontario chert sourcing including the discovery of Haldimand chert outcrops
(Moerschfelder 1985; Parker 1986a; Fox 2009). For the purposes of this study I accept Ellis’
(2004a: 61) assertion that the depicted Hi-Lo point reported by Roberts’ from the ENL 502 Hi-Lo site is actually of Haldimand chert. In addition, for the purposes of this study I consider that the other Ancaster chert points reported by Roberts (1985: 86) are in fact made from Haldimand chert. This decision was based on the arguments of Ellis (2004a) and the fact that Ancaster chert
(alternatively referred to as Goat Island chert) is rarely reported for Ontario Hi-Lo (but see
Tinkler and Pengelly 2004 and elsewhere Smith et al. 1998: 10) and the fact that Roberts’ identifications occurred prior to discovery of Haldimand chert outcrops (Moerschfelder 1985,
Parker 1986a; Fox 2009). This assumption has a significant impact on raw material usage patterns in the areas surveyed by Roberts. Raw material results of this study should be considered with this assumption in mind. Further work to conclusively determine raw material affiliation for these points is necessary to confirm raw material identifications and clarify the Hi-Lo record north of
Lake Ontario.
Five of Roberts’ points are depicted (1985: Fig. AI-la) without reference to site location along with a sixth (1985: Fig. AIV-2) which had been previously identified as Ancaster chert from the
ENL 502 (AlGo-36). Of the five Hi-Lo points depicted by Roberts (1985: Fig. AI-la) the points
3rd and 4th from the left on the top row are considered to be Haldimand chert based on their similarity to the ENL 502 point which has been later reclassified as Haldimand chert. The remaining three points to the right of the Haldimand Hi-Lo points are considered to be Onondaga.
These points are entered in to the 2015 Hi-Lo Sample without reference to provenience so as not to influence spatial patterning results. Only the point from ENL 502 (Roberts 1985, Fig. AIV-2) was included in the 2015 Hi-Lo Sample Distribution.
73
Six of Roberts’ (1978, 1979, 1985) Hi-Lo site locations, known through MTCS records, are included in the Expanded 2015 Hi-Lo Distribution. The status of the sites known through MTCS records within the 2015 Hi-Lo sample is shown by Table 3.
Table 3. Roberts (1985) Hi-Lo sites plotted in Fig 19. Ontario and western New York Hi-Lo distribution (expanded 2015 Hi-Lo sample)
Region Site Name Borden # Biface Included in Biface Raw 2015 Hi-Lo Sample Material Durham ENL 502 AlGo-36 Yes Haldimand (1) Durham Harold AlGq-25 No Unknown Stevens Durham West Whitby AlGr-26 No Unknown Townline Durham N/A AlGo-33 No Unknown Burlington- Marchesse AjGw-40 No Unknown Oakville Burlington- N/A AhGw-64 No Unknown Oakville
Tinkler and Pengelly (2004: Table 7.1) report 28 Hi-Lo points from the Niagara peninsula which were recorded by Jim Pengelly in various Niagara Peninsula collections. Summary raw material identifications are presented (2004: Table 7.1) as well as 19 plotted Hi-Lo finds (2004: Table
7.6b). These site locations have been included in the Expanded 2015 Hi-Lo Sample Distribution.
The Tinkler and Pengelly (2004) Hi-Lo sample raw material identifications and find distribution are summarized and depicted in Figure 7.
74
Figure 7. Summary of Hi-Lo documented in Tinkler and Pengelly 2004 showing site locations included in 2015 Hi-Lo Sample
Only two of the plotted Tinkler and Pengelly Hi-Lo finds have been tentatively identified and included in the 2015 Hi-Lo Sample Distribution. It was assumed that Cox I (AfGt-62) and Cox II
(AfGt-62), which are located very close to one another were represented as a single location by
Tinkler and Pengelly. The status of the identified sites within the 2015 Hi-Lo Sample are shown by Table 4. The remainder of the plotted sites cannot be confidently associated with any of the points mentioned by Tinkler and Pengelly (2004: Table 7.2). The additional raw material data provided by these points is included in the Expanded 2015 Hi-Lo Sample.
Table 4. Tinkler and Pengelly (2004) Hi-Lo included in 2015 Hi-Lo Sample
Region Site Name Borden # Biface Included in Biface Raw 2015 Hi-Lo Sample Material Niagara Cox I AfGt-62 Yes Onondaga (1) Niagara Cox II AfGt-63 Yes Onondaga (1) Niagara FEHM N/A Yes Haldimand (1)
75
Some additional points whose provenience and raw material affiliation was known through CRM reports, doctoral dissertations or the Ministry of Tourism, Culture and Sport’s records were also included in the Expanded 2015 Hi-Lo sample. Although the bifaces were not included in the 2015
Hi-Lo sample these points are of use in documenting raw material usage patterns. Details of these points and references are presented below in Table 5.
Table 5. Hi-Lo Sites plotted in Fig 19. Ontario and western New York Hi-Lo distribution (included in Expanded 2015 Hi-Lo sample) Region Site Name Borden # Included in 2015 Reference Raw Material Hi-Lo Sample Grand Martinello 5 AjHb-19 No MTCS Onondaga (1) River 2014 Drainage Grand Hilborn AjHb-21 No MTCS Kettle Point (1) River 2014 Drainage Grand Hanlon Field J AiHb-287 No MTCS Onondaga (1) River 2014 Drainage Grand Glass No Deller Haldimand (2) River 1988 Drainage
Some additional Ontario points attested in the literature (see Wright 1978) were not able to be examined due to the limitation of time on this research and the impossibility of recording the sheer quantity of Hi-Lo points in the province. These points include ‘several’ Kettle Point chert
Hi-Lo points in Haldimand county (Moerschfelder 1981, 1982) and other Kettle Point and
Haldimand chert specimens found in the area of Saint Thomas between the Haldimand and Kettle
Point chert outcrops (George Connoy pers comm cited in Parker 1986: 14). An additional Kettle
Point chert Hi-Lo point has also been observed in a local area collection originating in the
Binbrook area (Dan Long pers comm 2015). Turning north, the unexamined point from the
76
Priceless site (AiHb-227) (New Directions Archaeology Ltd. 2001) represents one of the very few
Hi-Lo points recovered thus far from Wellington County.
4.3.B Implications of Non-Projectile Functional Modifications of Form for Sample Selection
Previous work by Ellis and Deller (1982, 2014) supports the idea that Hi-Lo points were often used serve non-projectile point uses. While tip impact fractures occur (Figure 8), other points
(Figures 9 and 10) exhibit extensive beveling on a single lateral edge, the effects of which upon symmetry call into question their use as projectile tips and suggests a non-projectile function for these tools. Ellis and Deller (2014: 13) draw a distinction between "longitudinal" and "lateral" resharpening, “Longitudinal refers to reworking that results in changes in tip configuration and blade length while the lateral refers to reworking directed laterally which reduces the width of the
"original" point form and alters lateral edge configuration.” Figure 11 provides an example of longitudinal resharpening on a reworked point resulting in a blunted tip more suited for use as a scraper than a projectile.
Figure 8. Hi-Lo Figure 9. Hi-Lo displaying Figure 10. Hi-Lo Figure 11. Hi-Lo displaying tip extensive bevelling on a displaying with extensive impact. Note false single lateral edge “perforator” use longitudal flute resulting from possiblly due to modification (see Ellis resharpening. Note tip impact (Ellis and functional modification and Deller 2014: 10) blunted tip Deller 2014) for non-projectile use as suggesting use as a side scraper (Ellis and an end scraper Deller 2014 (Ellis and Deller 2014
77
Hi-Lo hafted biface forms are referred to in the text as “points” or “projectile points” without implying a membership in an exclusive function as projectile components. Formal variability in the blade element suggests that what are often loosely termed as Hi-Lo projectile points may often be more accurately described as multifunctional tools (Ellis and Deller 2014). Hi-Lo hafted biface forms whose morphology suggested primarily a non-projectile function are included in this analysis. This approach is consistent with the methodology of previous studies of Hi-Lo point variability undertaken by Ellis and Deller (1982) and White (2012).
4.4 2015 Hi-Lo Sample Provenience
The geographic provenience of each point was plotted on a map of Ontario using QGIS 2.4.0-
Chugiak software. The 2015 Hi-Lo Sample (n=302) is primarily composed of points whose provenience can be traced to a specific lot. The geographic provenience of points was plotted using concession, lot, and township information as well as site maps contained within published technical reports. If provenience was known to an area defined by adjacent lots, the point was assigned coordinates corresponding to the centre line between the two lots. UTM coordinates in this study were determined using North American Datum 1983 (NAD83) working in Zone 17T.
The sample includes a number of points whose provenience is known only to the township level
(n=5) and those known to the county level (n=2). For these points, UTM coordinates were assigned based on a position approximating the centre of the township or county.
There are also a number of points (n=28) whose provenience is not known beyond an affiliation with southern Ontario. These points were not assigned geographic provenience. No approximated provenience was assigned to these points, as doing so would have unduly influenced the aspects of this study concerned with spatial patterning in Hi-Lo variability.
78
4.5 Maximum Thickness
Maximum thickness (mm) was measured using calipers or drawn from published data if the collection was not available for analysis. The measurement was recorded at the point of maximum thickness along the axial plane. Points broken or points damaged in a way which made it unclear if the thickest portion had been removed were not included in analyses involving maximum thickness.
4.6 Weight
Point weight was recorded in person or drawn from published data where available. Since point weight may be expected to vary between different types of chert it is important to consider the raw material affiliation of points before drawing conclusions based on weight. Weight is treated here as a separate variable from morphology. Its patterning is examined independently by raw material.
4.7 Raw Material Identification
Raw material identifications were made based on macroscopic characteristics. When possible, raw material identifications were made through personal analysis of color, mottling, banding patterns, texture, luster, and inclusions. Examples of points identified as having been manufactured from Haldimand chert (Figure 12), Kettle Point chert (Figure 13), Onondaga chert
(Figure 14) and Bayport chert (Figure 15) are included below.
In cases where collections were not available for personal analysis, raw material identifications were drawn from published data. In cases where points were unavailable for personal analysis and raw material affiliation could not be determined from published descriptions or graphics, points were considered to be of unknown raw material affiliation unless otherwise specified.
79
Figure 12. Hi-Lo point of Figure 13. Hi-Lo point of Figure 14. Hi-Lo Point of Figure 15. Hi-Lo Point Haldimand chert, Stelco I Kettle Point chert, Leslie Onondaga chert, Cox II site of Bayport chert, site (AfHa-200) site (AhGw-32) (AfGt-63) Welke-Tonkonoh site (AfHj-5)
4.8 Landmark Placement for Geometric Morphometric Analysis of Hi-Lo Bifaces
All images were imported into the morphometric program tpsDIG2 (Rohlf 2008). On complete specimens, a total of fifteen landmarks were recorded (see Figure 16, Table 6). Points were included in the sample if they fit two criteria: (1) they could be confidently typed as Hi-Lo, (2) the majority of the landmarks designated for the haft element (LM 2-9) were present. Landmarks 1 –
12 were placed in homologous locations to ensure consistency of analysis between specimens
(Lycett 2009: 81). Landmarks 13 – 15 are considered to be mathematical landmarks (Barceló
2010). Semi landmarks 3 and 4 as well as 16 and 17 were applied on an as needed basis to capture the angle of contraction for bases which had no expansion of the ears.
80
Table 6. Landmark positions and descriptions Landmark ID Description 1 Tip 2 Point of maximum basal concavity 3 and 4 Point of maximum nasal constriction 5 and 6 Point of maximum basal width 7 and 8 Distal extremes of ears 9 and 10 Marking the beginning of significant basal constriction 11 and 12 Point of maximum width above the ears 13 Mid-point between LM 1 and 2 14 Marking the wdge of the point at 270 degree right angle from LM 13 15 Marking the edge of the point at 90 degree right angle from LM 13
16 and 17 Applied on as needed basis above LM 3, 4 and below LM 9 and 10 to capture contracting base angles
Landmark designations and measurements were focused on the area defined as the haft region which is defined as distal to landmarks 9 and 10 which marks the beginning of significant basal constriction. The haft region is assumed to have undergone the least amount of change in size and shape due to reworking after its initial manufacture (Ellis 2004; Ellis and Deller 2014; Thulman
2012; White 2013). It has also been suggested by Ellis (2004a) that Hi-Lo basal configurations may be of use in defining subtypes within the Hi-Lo typology which may have use as temporal markers. The utility of differences in basal configurations to discriminate typological classes has recently been underscored by Thulman (2012) who used geometric morphometrics to discriminate between Florida Paleoindian points classified as Suwannees, Simpsons, and Transitional Side
Notched.
81
Figure 16. Selection of metric variables included in analysis (Image not to scale)
4.9 Metric Variables
Landmark positions were chosen to allow for the investigation of Hi-Lo variability using three sets of analytical devices. Cartesian coordinates (mm) of each landmark were determined using tpsDIG2 software (Rohlf 2008). These values were used to calculate 12 linear variables and two angle measurements (Tables 7, 8 and 9).
82
Table 7. Metric Variables defined by horizontal differences
Variable Landmarks Description DEW 8-7 Width at distal extremes of ears NW 4-3 Width at point of maximum basal constriction (neck) BW 6-5 Width at point of maximum basal width (corner) SW 10-9 Width at points marking the beginning of basal constriction EE Mean((3-5),(6-4)) Mean amount of ear expansion BLW 12-11 Blade width TD ABS((Mean 14, 15)–1)) Blade straightness
Table 8. Metric variables defined by vertical differences
Variable Landmarks Description BCon 2-(7 or 8) Maximum offset of distal edge (depth of basal concavity) BCon.NW 2-(Mean 3, 4) Mean distance between maximum basal offset and point of maximum basal constriction BCon.BW 2-Mean 5, 6) Mean distance between maximum basal offset and point of maximum basal width NW.BW (Mean 3, 4)-(Mean 5, 6) Mean offset of maximum basal constriction from maximum basal width BLL 1-(Mean 9, 10) Blade length NH (Mean 3, 4)-(Mean 7, 8) Neck height HL (Mean 9, 10)-(Mean 7, 8) Haft length S.CL (Mean 9, 10)-(Mean 5, 6) Shoulder to corner length ML 1-(7 or 8) Distance between tip and distal extreme of base BLWH (Mean 11,12)-(7 or 8) Distance between point of maximum blade width to distal extreme of base
83
The first of these analytical sets is comprised of the traditional variables utilized in hafted biface analysis (Andrefsky 2005: Fig. 7.32). These variables include: (B) width at point of maximum basal constriction (i.e. neck width), (C) width at point of maximum basal width (i.e. base width),
(E) blade width, (K) blade length, (L) neck height, (M) haft length, (N) shoulder to corner length.
Landmarks and metric variables were also chosen to ensure compatibility of the dataset with the recent data generated by Andrew White (2012, 2013) in his study of biface variability in
Paleoindian and early Archaic point types in the American midcontinent. White’s study analyzed a wide range of projectile point types including Hi-Lo and Dalton. Hi-Lo points (n=192) were drawn from areas concentrated in northern Indiana, northern Ohio, southern Michigan, and southwestern Ontario (White 2012: Fig. 5). While there exists some overlap in the sample used by
White and this study, the samples are largely distinct. It was therefore thought that methodological consistency would produce the most productive results useful in larger scale studies. Landmarks were placed so as to allow for the analysis of the eight metric variables considered by White (2012: Fig 7.7). Some minor changes have been applied to the verbal definitions of these variables to maintain consistent terminology throughout the text and so, the mathematical definition of the metric variables remains constant.
4.10 Character State Analysis
Points were analyzed according to the character state analysis method utilized by Michael
O’Brien and colleagues (O’Brien et al. 2014) in their phylogenetic study of Paleoindian projectile points. This phylogenetic study utilized a character state approach to variability where measurements were coded following the paradigmatic classification system expressed by O’Brien et al. (2001). The character state analysis combines the use of metric variables (i.e. length/width ratio) with qualitative assessments of point attributes (i.e. base shape, tang tip shape). Graphic representations of the characters (Roman numerals) and character states (Arabic numerals) and their descriptions are presented below (Figure 18, Table 9).
84
Table 9. Character state analysis adapted from O’Brien at al. 2014 Character State Variable ID Description Max Blade BLWH Height of maximum blade width—quarter section in Width Height which the widest point of the blade occurs. Basal Shape - Overall base shape—qualitative assessment of the shape of the basal indentation. Basal BIR Basal indentation ratio—the ratio between the medial Indentation length of a specimen (1 – 2) and its maximum length; Ratio the smaller the ratio, the deeper the indentation. Constriction CR Constriction ratio—the ratio between the minimum Ratio blade width (proximal to the point of maximum blade width) and the maximum blade width; the smaller the ratio, the higher the amount of constriction. Mean Basal MBA The degree of tang expansion (or contraction) from the Angle long axis of a specimen. The lower the angle, the greater the expansion. Ear Tip Shape - Ear-tip shape—the shape of the tip ends of tangs. Length/Width L.W Length/width ratio—the maximum length of a specimen Ratio divided by its maximum width.
While O’Brien et al.’s (2014) phylogenetic study was principally concerned with fluted projectile points, it also included point types such as Dalton which may occasionally be fluted. For this reason, the presence or absence of fluting was considered as a presence-absence character
(O’Brien et al. 2014: Fig. 5, Character State VII). Fluted points were distinguished by the presence of what was termed by Bruce Bradley (1997) as a morphological flute. Morphological flutes are understood as “basal flake scars that extend past the point of the hafting element and are visible on the finished object.” (Bradley 1997: 54-55).
In the case of Hi-Lo, Ellis and Deller (2014: 5) interpret the presence of long, flute-like basal thinning flakes on points as accidental, noting that “there seems to have been no consistent attempt to remove those long flakes from the base but simply on occasion one carried a little
85 farther.” In agreement with the judgment of Ellis and Deller (2014), no deliberate morphological fluting was detected during the course of this analysis. Thus, the fluted vs. unfluted character state of O’Brien et al. (2014) was not included in this analysis as all Hi-Lo points included in the sample are considered to be unfluted
Character State V expresses the angle of the basal tangs (i.e. ears). For points with lateral basal constriction proximal to ear, angles express the degree of expansion for the ear (Landmarks 5, 6) from the point of maximum basal constriction (LM 3, 4). In cases where the base constricts continuously from the shoulders (LM 9, 10), landmarks 16 and 17 were used to calculate the basal angle. Angles on contracting bases were measured as greater than 90o in a manner expressed mathematically in Table 11 and graphically by Figure 17.
Table 10. Angle measurement - straight and expanding tangs (90 degrees or less)
Variable Landmarks Description O Degrees(ATAN(4y-6y)/(6x-4x)) Outer tang angle 1 P Degrees(ATAN(3y-5y)/(5x-3x)) Outer tang angle 2
Table 11. Angle measurement - contracting tangs (< 90 Degrees)
Variable Landmarks Description O (ABS(Degrees(ATAN(17y-4y)/(17x-4x)) + 90o Outer tang angle 1 P (ABS(Degrees(ATAN(16y-3y)/(3x-16x)) + 90O Outer tang angle 2
Figure 17. Contracting ear angle measurement (P)
86
87
Figure 18. Character state variable adapted from O’Brien et al. 2014
88
Chapter 5
Results and Analysis
5 Results and Analysis
The purpose of this chapter is to describe the recorded variability of a sample of Hi-Lo type projectile points from southern Ontario and New York. A total of 302 points form the sample for this study; 291 from Ontario and 11 from New York. Spatial statistics are applied to the data derived from the variables explained in Chapter 4 in an attempt to test for the existence of geographically defined point sub-styles within the Hi-Lo type. The distribution of Hi-Lo points within the study area is described in terms of implications for our understanding of Hi-Lo raw material usage patterns and geographic patterning. The influence of differential knapping qualities between the main chert types used by Hi-Lo knappers is explored as a means to understand the significance of raw material constraints as a source of variability within the Hi-Lo type.
5.1 Type Description
Aspects of Hi-Lo base shape which, are considered indicative of the type include; concave bases, short hafts, and variable lateral haft margins (Fitting 1963; Justice 1987; Ellis 2004a; Ellis and
Deller 1982, 2014; White 2013). Lateral haft configurations vary with stemmed, shallowly side- notched and un-stemmed lanceolate forms being included within the Hi-Lo type (Ellis 2004a).
Basal ends are either slightly concave to moderately concave (Ellis and Deller 2014: 5). Points typically display thick, broad ears which are highly variable (Ellis and Deller 2014: 5). Ear configurations are often asymmetrical with tips most commonly blunt or rounded, but in rare cases pointed. Grinding on the haft element is common but not ubiquitous. The degree of basal grinding is variable; some points are only lightly ground while others may display quite dulled basal edges due to significant basal grinding. While basal grinding is most common on fully
89 formed points it is not restricted to finished points. The degree of grinding and its extent varies between points, some points may be disproportionally ground in the basal concavity or the lateral haft margins. This grinding may be a modification to mitigate torqueing stresses during cutting activities. Blade shape is highly variable (see Fitting 1975; Ellis and Deller 1982, 2014) and is not the best identifier of Hi-Lo. It is not the purpose of this section to re-describe the Hi-Lo type, for in-depth definition and discussion the reader is directed to the recent work of Ellis and Deller
(2014).
5.2 Ontario and Western New York Hi-Lo Distribution
At the present time, in Ontario, definitive Hi-Lo points have been only been recovered from the southern portion of the province. The known distribution of Hi-Lo points in southern Ontario and western New York is displayed in Figure 19, while the distribution of points included in the present study is depicted by Figure 20. Hi-Lo points have been found in an area bounded in the north by the Frontenac Axis of the Canadian Shield east of Lake Simcoe, on the north-west by a line running southwest roughly from the southern edge of Lake Simcoe to Stratford and from there west to Lake Huron in the area of Grand Bend.
Figure 19. Ontario and western New York Hi-Lo distribution (Expanded 2015 Hi-Lo Sample distribution)
90
Figure 20. 2015 Hi-Lo sample distribution
5.2.A Possible Occurrences of Hi-Lo in Eastern Ontario Some points whose form bears resemblance to Hi-Lo have been noted from further afield in eastern Ontario, however at the present time the typological association and relationship of these points to the main southern Ontario distribution is unclear. A compelling example, depicted in
Figure 21, is drawn from an assemblage classified as late Paleoindian from Grenadier Island near the beginning of St. Lawrence River. Additionally, the Green site (BdGb-2) (Watson 2001;
Wright and Watson 1998, 1999) and another possible occurrence in the township of Tay Valley,
Lanark County (ARA 2010) may indicate the presence of Hi-Lo in eastern Ontario.
Figure 21. Grenadier Island biface
91
5.2.B Western New York Hi-Lo Distribution Hi-Lo points are rarely reported in New York. Most reported points cluster in western New York in the area of the Niagara Frontier (Smith et al. 1998, 2010; Houston and Perelli 2009) although
Hi-Lo points have been found as far east as Steuben County at the Kilmer site (Tankersley et al.
1996). The apparent scarcity of Hi-Lo points in New York may in part be related to the prominence of taxonomic systems and past interpretations of paleo-environment in the Northeast which downplayed the potential for late Paleoindian archaeology in the area. Support for this position is furnished by the fact that reported occurrences of Hi-Lo decrease dramatically at the modern Canada-USA border at the Niagara River (see Fig. 1). Ritchie’s (1961, 1971) influential typological scheme for New York State projectile points contained no reference to the Hi-Lo type which would be defined in by James Fitting in Michigan in 1963. The perceived dearth of late
Paleoindian and early Archaic material (Ritchie 1969, 1979; Fitting 1968, 1975; Ritchie and Funk
1971, 1973) in the Northeast, was seen as a result of paleo-environmental conditions in the early
Holocene Northeast leading to an inhospitable, low productivity forest incapable of supporting substantial human populations. This model implied an early Holocene occupational hiatus in the
Northeast and Great Lakes regions. The occupational hiatus was explained by the "Ritchie-Fitting
Hypothesis" which held "that the extreme scarcity in the Northeast of Late Paleo-Indian [and
Early Archaic] traces arises primarily from the fact that around 7,000 B.C. the forest composition was undergoing a marked alteration from spruce-pine to pine with a significantly lowered carrying capacity for game" (Ritchie and Funk 1971: 46).
The predominance of Richie’s typological system in New York may have led to points conforming to the Hi-Lo type being attributed to other typological classes, a classificatory issue which could in part continue to this day. Documenting the occurrence and frequency of Hi-Lo points in New York is an important avenue of research which serves to underscore the influence of modern typological systems in archaeological interpretation.
92
5.3 Raw Material Identification and Distribution
The present study uses a sample of 302 Hi-Lo points distributed across the southern portion of
Ontario and western New York. Summary raw material identifications for the Expanded 2015 Hi-
Lo sample are illustrated by Table 12. Raw material identifications for the 2015 Hi-Lo sample are shown by Table 13.
Previous investigations of Ontario Hi-Lo raw material usage patterns have shown Haldimand,
Onondaga and Kettle Point chert to be the primary sources of toolstone used by Hi-Lo knappers
(Ellis and Deller 1982, 2014; Roberts 1985; Tinkler and Pengelly 2004). These studies have focused on specific areas within southern Ontario and provide some idea of regional differences in Hi-Lo raw material usage patterns. Figures 22-24 depict the distribution of Hi-Lo finds attributed to the major raw material groups; Haldimand, Kettle Point and Onondaga chert.
Table 12. Raw material affiliation (Expanded 2015 Hi-Lo Sample) Material Type n % Bayport 8 2.29 Haldimand 155 44.29 Kettle Point 50 14.29 Onondaga 79 22.57 Non-chert Material 1 0.29 Selkirk 1 0.29 Trent 1 0.29 Ancaster 6 1.71 Mercer 1 0.29 Quartz 1 0.29 Unidentifiable/Not Available 47 13.43
Table 13. Raw material affiliation (2015 Hi-Lo Sample) Material Type n % Bayport 8 2.65 Haldimand 133 44.04 Kettle Point 49 16.23 Onondaga 61 20.20 Non-chert Material 1 0.33 Selkirk 1 0.33 Trent 1 0.33 Ancaster 1 0.33 Unidentifiable/Not Available 47 15.56
93
Each of the major raw material groups are distributed across a wide swath of southern Ontario.
However, due to the limited sample size some parts of the southern Ontario remain underrepresented (e.g. Essex County). These data provide the most comprehensive look at the material used by Hi-Lo knappers for biface production in Ontario and western New York.
These data can be discussed in relation to Ellis’ (2004a) observations on Hi-Lo lithic procurement. Ellis (2004a: 59) noted that all Hi-Lo artifacts “… are made on fine-grained materials such as cherts” and that “If one relies solely on data related to the raw material used for points in southern Ontario, then Hi-Lo knappers relied heavily on Haldimand and Kettle Point cherts and less so on other cherts such as Onondaga, Selkirk and Bayport (Deller 1989).” Note that the second statement makes no mention of Lockport-Ancaster-Goat Island chert use by Hi-Lo knappers.
Figure 22. Haldimand chert Hi-Lo distribution
94
Figure 23. Onondaga chert Hi-Lo distribution
Figure 24. Kettle Point chert Hi-Lo distribution
Onondaga chert appears to have been more central to Hi-Lo lithic procurement than previously thought. Initial conceptions of Hi-Lo lithic usage patterns in Ontario were conceived using collections primarily drawn from Middlesex County in relatively close geographic vicinity to the
Kettle Point chert outcrop (Ellis and Deller 1982). The proximity to Kettle Point chert likely introduced some degree of bias into our understanding of chert use for Hi-Lo point production.
Additional raw material data for points in the Expanded 2015 Hi-Lo sample acts to partially correct this geographic bias by considering points from areas in the vicinity of Haldimand and
95
Onondaga chert outcrops as well. These additional points significantly alter the sample available for interpreting raw material selection for Hi-Lo point production. It now seems that if one relies solely on data relating to the raw material used for points in southern Ontario, then Hi-Lo knappers relied heavily on Haldimand, Onondaga and Kettle Point cherts and less so on cherts such as Selkirk, Bayport and Trent.
Turning to the proposition that Hi-Lo artifacts are exclusively made of chert, from the Seaton
Lands, there are some non-chert examples in the sample. For example, the Block H (Findspot 2)
Hi-Lo point has been described as “heavily weathered and appears to have been made from a heavy-grained non-chert material” (ARA 2005). The Seaton Lands point bears some resemblance to the projectile point from the Green Site (BdGb-2) near Smith’s Falls, Ontario which is also made of a non-chert material (Watson 2001; Wright and Watson 1998, 1999). Although the sample size is exceedingly small at this time, it is tempting to suggest that these two points may represent a distinct northerly expression of Hi-Lo.
Figure 25. Seaton Lands - Block H (Findspot 2) (ID-196)
96
Some questions remain about the use of Ancaster chert by Hi-Lo knappers. While Ellis (2004a:
61) is correct in noting that Robert’s (1985: 86, 211-212) raw material identifications were made without the knowledge of Haldimand chert the same cannot be said for identifications made by
Jack Holland (Smith et al. 1998: 8; Tinkler and Pengelly 2004). Holland made use of macroscopic and low-level (5-40x) incident-light microscopic examinations with reference to outcrop collected reference material for raw material identification. Cherts of the Lockport formation were formed on sponges and include spicules which are clearly visible through microscopic inspection (Laird
1935: 254–55; Wray 1948: 37; Fox 2009) and thus are likely able to be definitively identified.
The question of contrary identifications of Ancaster chert may be easily resolvable with reanalysis of relevant collections. This question is important to clarify Hi-Lo raw material usage patterns. If
Roberts’ (1985) Hi-Lo points are in fact Ancaster chert then they, together with points known through work of Tinkler and Pengelly (2004) and Smith et al. (1998), would indicate a geographically limited area of Ancaster chert usage around western Lake Ontario. Unfortunately, no Hi-Lo points previously attributed to Ancaster chert could be assessed personally during the course of this research.
Only one Bayport chert Hi-Lo point which had not previously been attested in the literature was noted during the course of this research. The point was found in Elgin County in the area of St.
Thomas (Figure 26). Bayport chert remains an exceptionally rare raw material for Hi-Lo points in
Ontario. This material is especially interesting as it suggests either long distance mobility for direct toolstone acquisition or exchange with peoples living closer to Bayport chert sources in
Michigan. During at time of the Lake Stanley low stand in the Huron Basin (c.a. 10,000 BP), it was possible to walk from Lambton/Middlesex County in Ontario to the Clarity Island hill in the
Lower Saginaw Valley in Michigan where Bayport chert is abundant (McCarthy et al. 2015). This raises the question of why Hi-Lo points made from Bayport are so rarely found in Ontario. The answer may relate to social factors which are extremely difficult to reconstruct given the nature of
97 the archaeological evidence available to researchers. Exploratory work documenting Hi-Lo raw material usage patterns in the area of the modern terrestrial border between Ontario and Michigan is necessary to begin to understand the factors which limited the spread of Bayport chert into
Ontario.
Figure 26. Bayport chert Hi-Lo - Elgin County, Ontario (ID-292) 5.4 Regional Raw Material Analysis As part of the assessment of spatial patterning in Hi-Lo variability the study area was subdivided into five distinct subregions. While these regions are in many ways arbitary they act as necessary heuristic devices enabling dicussion of regional variability in Hi-Lo size and raw material usage patterns. These geographic subregions will also be used to test for non-random spatial patterning in metric variability. Boundaries of these subregions can be seen in Figure 27. A summary of Hi-
Lo finds for each subregion by raw material type is provided for the Expanded 2015 Hi-Lo sample in Figure 27 and for the 2015 Hi-Lo sample in Figure 28.
98
(1) The Middlesex County and Southwestern Ontario subregion includes the Caradoc Sand Plains
(see Chapman and Putnam 1984: 113, 146) and important sites such as Welke-Tonkonoh which have previously provided samples for other Hi-Lo studies (e.g. Ellis and Deller 1982, 2014).
Points in this region are principally drawn from the Caradoc Sand Plains with smaller quantities coming from other locations in Middlesex, Lambton, Chatham-Kent and Essex counties. Drainage for this region is provided by the Ausable, Sydenham and Thames River systems (Chapman and
Putnam 1984). This subregion is also notable for including the outcrop of Kettle Point chert (Fox
2009; Janusas 1984).
(2) The Erie Lowlands subregion is roughly composed of areas of the Ekfrid Clay Plain and
Norfolk Sand Plain east of the St. Thomas moraine (see Chaptman and Putnam 1984: 113). This region is bounded on the east by the beginning of the Grand River drainage. It falls between the
Thames River drainage and the Grand River drainage and occupies a position in between the principal bedrock outcrops of Kettle Point chert in the west and Haldimand and Onondaga chert in the east (Fox 2009). Hi-Lo points in the Erie Lowlands subregion are not particularly well represented in academic or CRM literature. Despite being home to some well published sites (e.g.
Stelco I – Timmins 1995) this region has relatively few documented Hi-Lo finds. A large body of the Hi-Lo points from this region are known through an exploratory survey by the author, Steven
Timmermans and Ramsay MacFie of farm collections in Malahide and Bayham townships during
July 2015.
(3) The Grand River Drainage subregion encompases areas to the east and west of the Grand
River. This region is bordered approximately on the east by the Niagara Peninsula and on the west by the Norfolk Sand Plain (Chapman and Putnam 1984). This region importantly includes outcrops of Haldimand chert and notably the Allan site, a Haldimand chert quarry site which has yielded a side-notched Hi-Lo point (Parker 1986a, 1986b). Other prominent sites include the
Snow Hill and Double Take sites (Ellis et al. 2009; Dickson 2011; TMHC 2004).
99
(4) The Niagara and New York subregion encompases a large and loosely defined geographic area. Points in the region have mostly been found along the Welland River in the Niagara
Peninsula and below the Onondaga Escarpment near Lake Erie (Fig. 8). Points from the United
States are drawn from the general Niagara Frontier area (Smith et al. 1998), although the Kilmer site (Tankersley et al. 1996) outlier pushes the boundaries of this region southeast to the
Appalachian Uplands. This subregion is home to numerous outcrops of high quality Onondaga chert which have played an important roles in the Ontario and New York prehistory (Fox 2009).
(5) The North of Lake Ontario subregion is encompases all points found to the east of the Niagara
Escarpment and north of Lake Ontario. This region is bounded arbitarily in the northeast by the
Frontenac Axis of the Canadian Shield. The North of Lake Ontario subregion is the only subregion where the majority of points have been found in the Canadian Biotic Province (see Ellis et al. 2009: Figure 22.6). Hi-Lo points in this region have typically been found south of the Oak
Ridges Moraine along the shore of Lake Ontario (see Roberts 1985), although increasingly finds are being reported further north in the Plainsville area (Jackson 2004).
100
Lo Sample raw material by geographic region for points with known provenience n=335 region provenience with points for geographic known bySample Lo raw material
- Figure27. Expanded Hi 2015
101
0
w material by geographic region for points with known provenience n=28 proveniencewith points known region material for geographic by w
Lo Sample ra Sample Lo
-
. 2015 Hi 2015 .
28
Figure
102
5.5 Metric Variable Analysis This section presents the results of the landmark geometric morphometrics analysis described in
Chapter 4. Basic descriptive statistics for the metric variables and character states for the total
2015 Hi-Lo sample and subgroups composed of points attributable to major raw material groups are presented in Tables 14-17.
Table 14. Summary descriptive statistics for 2015 Hi-Lo Sample (n=302)
Variable n Mean Minimum Maximum Standard Deviation W 226 7.68 1.50 21.60 3.06 T 243 7.92 4.80 13.10 1.20 DEW 272 15.00 7.75 24.35 2.98 NW 290 19.08 10.89 27.17 2.91 BW 266 20.22 10.89 29.99 3.01 SW 289 23.67 16.22 34.75 3.16 EE 268 0.59 0.00 2.68 0.44 BLW 294 24.30 16.57 36.49 3.33 BC 289 2.25 0.29 6.18 1.06 TD 264 1.17 0.00 5.31 0.98 BCon 299 2.98 0.37 7.45 1.12 BC.N 290 -2.71 -8.83 4.12 2.27 BC.BW 265 0.05 -5.26 4.84 1.36 N.BW 267 2.78 0.00 8.21 1.68 BLL 254 28.60 13.51 58.06 7.67 NH 267 5.26 0.09 14.09 2.18 HL 266 12.06 5.59 24.93 2.66 SCL 261 9.55 4.00 18.75 2.33 ML 265 41.23 24.39 78.89 8.73 BLWH 292 16.31 6.90 43.61 5.01 BIR 265 0.91 0.67 1.02 0.05 CR 282 0.79 0.48 0.97 0.09 MBA 303 81.57 61.10 125.96 8.10 L/W 261 1.71 1.10 2.98 0.32
Table 15. Summary descriptive statistics for Haldimand Hi-Lo Sample (n=133)
Variable n Mean Minimum Maximum Standard Deviation W 100 7.80 1.50 17.00 3.01 T 108 7.97 5.40 13.10 1.30 DEW 115 15.15 8.67 22.51 2.57 NW 124 19.39 11.76 26.00 2.80 BW 116 20.33 11.76 26.72 2.87 SW 123 23.73 16.70 34.75 3.13
103
EE 118 0.57 0.00 2.68 0.44 BLW 126 24.41 16.66 35.09 3.26 BC 124 2.13 0.29 6.18 1.12 TD 112 1.18 0.00 4.77 1.04 BCon 130 3.02 0.56 7.45 1.07 BC.N 124 -2.87 -8.83 3.40 2.34 BC.BW 116 -0.01 -4.15 3.42 1.32 N.BW 117 2.78 0.00 8.21 1.76 BLL 106 28.85 13.51 52.24 7.62 NH 114 5.39 1.08 14.09 2.23 HL 111 12.07 7.20 20.64 2.29 SCL 112 9.58 4.78 18.75 2.23 ML 114 42.06 24.39 78.89 9.33 BLWH 126 16.87 7.46 43.61 5.23 BIR 114 0.91 0.72 0.98 0.05 CR 119 0.80 0.49 0.97 0.09 MBA 133 82.33 61.75 125.96 8.80 L/W 111 1.74 1.10 2.98 0.37
Table 16. Summary descriptive statistics for Kettle Point Hi-Lo Sample (n=49)
Variable n Mean Minimum Maximum Standard Deviation W 45 7.86 2.60 21.60 3.53 T 48 8.00 6.20 11.10 1.10 DEW 46 14.60 7.84 23.74 3.20 NW 48 18.42 11.55 23.41 2.90 BW 44 19.84 12.98 26.55 2.87 SW 49 23.49 16.22 32.17 3.47 EE 44 0.65 0.00 1.87 0.41 BLW 48 24.14 16.57 34.44 3.68 BC 47 2.51 0.41 5.59 1.11 TD 43 1.24 0.08 5.31 1.03 BCon 49 3.01 0.90 5.16 1.03 BCon.N 48 -2.64 -7.37 1.84 1.95 BCon.BW 44 0.06 -5.26 2.64 1.26 N.BW 44 2.71 0.00 6.56 1.45 BLL 43 26.96 14.33 42.54 6.72 NH 43 5.64 2.37 9.58 1.89 HL 46 12.40 6.87 19.84 2.71 SCL 44 9.75 5.43 16.98 2.24 ML 43 39.49 24.79 57.35 7.55 BLWH 48 15.89 9.20 28.46 4.36 BIR 43 0.92 0.82 0.98 0.03 CR 47 0.76 0.55 0.96 0.09 MBA 49 78.35 61.10 101.82 6.90 L/W 43 1.66 1.25 2.16 0.24
104
Table 17. Summary descriptive statistics for Onondaga Hi-Lo Sample (n=62)
Variable n Mean Minimum Maximum Standard Deviation W 43 7.77 1.80 12.90 2.70 T 48 7.88 4.80 10.20 1.09 DEW 56 15.12 9.00 24.35 3.39 NW 61 18.92 10.89 27.17 3.08 BW 52 20.35 10.89 29.99 3.50 SW 60 23.98 19.02 31.99 2.95 EE 53 0.58 0.00 2.19 0.46 BLW 61 24.65 19.97 36.49 3.14 BC 61 2.40 0.70 5.09 0.98 TD 55 1.15 0.00 3.60 0.90 BCon 61 2.98 1.11 6.84 1.20 BCon.N 61 -2.59 -7.08 4.06 2.29 BCon.BW 51 0.17 -3.34 4.84 1.28 N.BW 52 2.82 0.00 6.87 1.80 BLL 53 28.84 16.65 47.04 6.95 NH 56 4.73 0.09 10.46 2.21 HL 54 11.96 6.18 19.94 2.66 SCL 52 9.64 5.97 15.82 2.31 ML 55 41.11 29.49 63.83 7.43 BLWH 61 16.30 6.90 31.66 5.06 BIR 56 0.91 0.69 0.98 0.06 CR 60 0.77 0.48 0.93 0.08 MBA 62 81.47 63.43 103.00 8.19 L/W 55 1.68 1.26 2.60 0.27
5.4.A Variable Correlation
To determine whether or not significant relationships exist between any of the measured variables correlation coefficients were calculated. The correlation coefficient quantitatively expresses the degree of correlation between two variables and it can be used to describe this relationship.
Postive values indicate a positive correlation where increases in values for one variable will be accompanied by increases in the other. Negative values signifiy an inverse relationship where increase in one variable will tend to be accompanied by decreases in the other. Correlation coefficients are a useful tool to undertand the relationships between different features of Hi-Lo points. Significant correlation coefficients (i.e., with >.8 or <-.8) are presented below in Table 18.
105
Table 18. Correlation coefficients of a magnitude >0.80
Variable NW SW BC BLL BW 0.92 BLW 0.95 ML 0.83 CR -0.85
Of the four significant correlation coefficients only the relationship between the constriction ratio
(CR) and the linear measurement of basal constriction (BC) can be described as inverse. This relationship emerges because lesser values for the CR are associated with points which have more relative difference between NW and the BLW. Since blade width (BLW) and shoulder width
(SW) are strongly positively correlated (see Table 18), increases in blade width will tend to be matched by corresponding increases in shoulder width. Therefore, any point with a slight neck and greater blade and shoulder width will return a high BC value and a lower CR value. The remaining significant correlation coefficients describe postive relationships. These relationships are largely intuitive (e.g. ML and BLL) and do not warrant much further explaination.
5.6 Effects of Raw Material on Hi-Lo Design
A potential source of variability in Hi-Lo design is related to differences in raw material quality and workability. While differences in raw material workability are difficult to assess quantitatively, experiential accounts of experienced modern flint knappers serve to shed some light on variable knapping properties (e.g. Long 2004). This section addresses the potential impact of raw material constraints as a driver of variability in Hi-Lo design. If raw material appears to not be play a significant role in determining Hi-Lo variation it can be excluded as a determinant variable in explaining the observed morphological differences within the biface sample. The impact of raw material on Hi-Lo variabiliy is addressed through a comparison of the coefficients of variation for each variable between the major raw material subgroups. A Mann-Whitney U test was used determine whether the variable data differs in statistically significant ways between the
106 raw material subgroups. Additionally, a principal components analysis was used to determine whether or not the spectrums of variability which primarily distingush Hi-Lo points display patterning showing substantial differences between raw material groups.
5.6.A Coefficients of Variation for Variables by Raw Material A coefficient of variation (CV) was calculated for each variable in the overall 2015 Hi-Lo sample and for points subgroups within the sample identified as being made from either Haldimand,
Kettle Point or Onondaga chert. Coefficients of variation are provided in Table 19 and contrasted graphically in Figure 29.
The coefficient of variation is defined as the ratio of the standard deviation to the mean. The coefficient of variation expresses the amount of variability in a set of values relative to its mean.
Because it not expressed in terms of measurement units the coefficient of variation can be used to compare the relative spread of the variability for variables with different means or those measured in different units (i.e. NW versus MBA). Since coefficients of variation cannot be meaningfully calculated on variables whose values may be negative, the variables BCon.N and BCon.BW were excluded from this analysis (as per White 2013: 91). Comparison of the coefficients of variation for the three well represented raw material groups will reveal patterning in the levels of variation for aspects of the Hi-Lo form captured by the variables. These data will determine if points attributable to the major raw material types within the 2015 Hi-Lo sample are characterized by differential patterns in the spread of their variability.
107
Table 19. Coefficients of variation for 2015 Hi-Lo Sample (n=302)
Variable Haldimand Kettle Point Onondaga 2015 Hi-Lo Sample W 38.66 44.85 34.75 39.89 T 16.30 13.70 13.79 15.20 DEW 16.94 21.92 22.42 19.89 NW 14.42 15.72 16.26 15.27 BW 14.09 14.45 17.18 14.86 SW 13.20 14.77 12.32 13.34 EE 77.35 62.90 80.25 74.98 BLW 13.34 15.24 12.75 13.72 BC 52.49 44.32 40.73 47.32 TD 88.46 83.07 78.19 83.97 BCon 35.37 34.36 40.33 37.66 N.BW 63.36 53.37 63.72 60.32 BLL 26.40 24.92 24.10 26.82 NH 41.32 33.45 46.75 41.50 HL 18.96 21.88 22.21 22.03 SCL 23.30 22.96 23.96 24.36 ML 22.19 19.11 18.07 21.18 BLWH 31.02 27.41 31.02 30.73 BIR 5.53 3.80 6.52 5.68 CR 11.13 12.22 10.82 11.17 MBA 10.68 8.81 10.05 9.93 L/W 21.12 14.55 15.82 18.87
100 90 80 70 60 50 40 30 20 10 0
Haldimand Kettle Point Onondaga
Figure 29. Coefficient of variation for Hi-Lo variables by raw material group
Coefficients of variation exhibit highly dissimilar amounts of variation with CV values for the overall 2015 Hi-Lo sample ranging from a low of 5.68 for BIR to a maximum of 83.97 for TD.
108
The variables for each of the raw material groups can be broken down into two main groups: (1) those with relatively low CV values (<25) for two of the three raw material groups; (2) those with
CVs deemed high (>25) for at least two of the three raw material groups (Table 20).
Table 20. Variable coefficients of variation showing significant differences between raw material groups
Variable Class Group 1 (<25) Group 2 (>25) Horizontal DEW, NW, BW, SW, BLW EE, BC, TD
Vertical BLL, HL, S.CL, ML BCon, N.BW, NW, BLWH Angle MBA Ratio BIR, CR, L.W Other T W
Variables in the second group tend to exhibit the greatest amount of differential variation between raw material subgroups. CV for variables in the first group typically differ between raw material subgroups only slightly (>5) while those in the second group tend to exhibit a larger degree of variation between raw material.
Variables that have a low CV for all three raw material groups tend to describe basic aspects of horizontal and vertical size (e.g. DEW, NW, BW, SW and HL, S.CL, ML). The low CV values for variables relating to aspects of the hafting element implies that points from each raw material group were made using a consistent design template meant to facilitate hafting in a standardized shaft (White 2013: 92). This line of reasoning can also be applied to the low values of the CVs for maximum point thickness (T). Points which were either too thick or too thin may not have been able to perform their desired function satisfactorily. The rather tight spread of variability expressed by the CVs for T imply that a fairly standard thickness was also a part of the Hi-Lo design template.
The weight (W) variable is of interest as it shows a substantial amount of variability between the three raw material groups. However, this variable is likely the most sensitive to innate differences
109 between Haldimand, Kettle Point and Onondaga cherts (see Fox 2009; Janusas 1984; Parkins
1977; Parker 1986a, 1986b). Additionally, any interpretation of the CV for this variable is somewhat confounded by the fact that points in the sample are drawn from different stages of a variable range of Hi-Lo reduction strategies (see Ellis and Deller 2014), a factor which has important implications for weight given the degree of resharpening common on Hi-Lo blades.
This issue is especially problematic for sensitive measurements like TD. Differential blade resharpening strategies will likely have altered blade shape and orientation significantly during the course of the artifact’s use-life. While this explanation accounts for the high CV for TD it also negates the variable’s use for determining the effects of raw material on Hi-Lo design. Since variability in TD is most likely driven by functional concerns, variability in TD may often result from design considerations related to immediate concerns. While variability in TD may also relate to initial tool design, the high likelihood that variability is a reflection of responses to functional concerns confounds its use as an indicator of raw material’s effects on initial design.
The greatest variability in CV between raw material groups is found in the ear expansion (EE) variable. While Onondaga and Haldimand chert exhibit similarly high CV values the value for
Kettle Point is substantially lower. To put this in context, a comparison of the mean CV for all variables of each raw material subgroup indicates that on the whole Kettle Point chert Hi-Lo points show slightly less observed variability than Onondaga and Haldimand (27.63, 29.18 and
29.80 respectively). Returning to EE, in absolute terms, Kettle Point chert Hi-Lo points exhibit the greatest degree of linear horizontal ear expansion compared Onondaga and Haldimand chert
(0.65, 0.58 and 0.57 mm respectively). While these numbers may seem minute they are worth considering further in relation to other relevant variables. Kettle Point chert Hi-Lo points also have the lowest observed mean basal angle (mean 78.35°) when compared to those of Onondaga chert (mean 81.47°) and Haldimand chert (mean 82.33°) (see further discussion in Section 5.5.3).
The more acute the angle, the more defined the portion of the basal margin marking the basal ear
110 expansion will tend to be. Haldimand, Kettle Point and Onondaga chert are all characterized by low CV values (Table 20) for MBA indicating that angles for each group are tightly clustered around their respective means. Kettle Point chert Hi-Lo points have an especially low CV for
MBA meaning that Kettle Point Hi-Lo points have both the lowest mean basal angle and the tightest spread of variation around that mean. When the CV for the EE variable is considered in light of the other relevant variables it appears that the ears of Kettle Point chert Hi-Lo points are in fact different from than those points made from of Haldimand or Onondaga chert (see further discussion in Section 5.5.3).
Variables N.BW and NH are closely related and are best interpreted as a pair. Both variables are dependent on the height of the point of maximum basal constriction (NW). This feature’s vertical location relative to the distal extreme of the base is highly dependent on the configuration of the base itself. In cases where the haft was either stemmed or constricting, the point of maximum basal constriction will tend to closely approximate the distal extreme of the ears. Conversely, when the ears expanded (EE= >0 and MBA < 90°) the point of maximum basal constriction occurs farther from the distal extremes of the ears. Since the location of the point of maximum basal constriction is essentially defined by the basal ear configuration both N.BW and NH should be considered in relation to variables EE and MBA. Here a similar pattern can be observed between the CV values for each raw material group for variables EE, N.BW and NH.
The horizontal linear measurement of basal constriction (BC) shows a substantial amount of variability between Haldimand, Kettle Point and Onondaga chert Hi-Lo points. This result is interesting as Haldimand chert Hi-Lo points were also observed to have the lowest mean BC in absolute terms followed by those made from Onondaga and Kettle Point chert (2.13, 2.40 and
2.51mm respectively). CV values indicate a more dispersed range of variability around the mean for Haldimand chert Hi-Lo points when compared to Kettle Point and Onondaga chert (52.49,
44.32 and 40.73 respectively). This variable is an interesting counterpart to EE. While EE tracks
111 the horizontal expansion of the ear from the point of maximum basal constriction, BC relates to the constriction from the shoulders to the point of maximum basal constriction. In many cases this variable will be significantly influenced by blade resharpening. Often Hi-Lo blades are resharpened in a way which leaves a defined shoulder above the haft but in many cases no evidence of shouldering remains as it has been removed during resharpening (see Ellis and Deller
1982; Ellis 2004a). While the point of maximum basal constriction may be expected to remain relatively constant over the course of the point’s use-life, shoulder configuration is more likely to change. Variability in BC is discussed further in Section 5.53.
The process of blade resharpening is also responsible for the high CV values for BLWH. While some questions remain about the definition of a ‘finished’ Hi-Lo point (see discussion in Ellis and
Deller 2014), the general location of the point of maximum blade width (BLW) can usually be understood as occurring either at the same point as the shoulders on more triangular blades or further up along the margins blade margins for points with convex blades. An important aspect of the relationship between these two variables is expressed by the strong positive correlation between SW and BLW (see Table 18). In cases where the location of BLW occurs at the same or similar location as SW this concurrence is likely due to blade reshaping. Thus, variability in
BLWH is best understood as driven by a mixture of functional concerns and initial tool design.
The variability in BCon most likely represents pure stylistic variability in that it is least likely to have changed over the course of the point’s uselife. Since BCon is measured as the distance between the point of maximum basal concavity and the ear with most extreme distal termination, this variable may be a good indicator of variability in initial tool design. Although Hi-Lo basal concavities may be less likely to change over the course of the point’s uselife, variability will in part be related to alterations made during the repair of broken ears (Ellis and Deller 1982: 10).
This variable is interesting because Onondaga chert points show the most dispersed range of variability around the mean when compared to Haldimand or Kettle Point chert (40.33, 35.37 and
112 respectively 34.36). That Kettle Point chert Hi-Lo points have the most tightly clustered variability is consistent with the general pattern of greater uniformity among Kettle Point chert specimens. It is interesting that Onondaga chert Hi-Lo points show the most dispersed range of variability despite the fact that the Onondaga Hi-Lo sample is roughly half the size of the
Haldimand sample.
5.6.B Mann-Whitney U Test Analysis In order to determine whether variability in raw material constraints has a significant effect on Hi-
Lo design a Mann-Whitney U test was performed on the all point data recorded. This analysis sought to assess the degree to which raw material differences accounted for the observed variability in Hi-Lo form. The Mann-Whitney U test is a nonparametric test which is capable of assessing whether or not two samples are derived from the same population. In this case, the null hypothesis holds that Hi-Lo points made from different raw materials do not differ significantly meaning that raw material plays no significant role in determining Hi-Lo form. The alternate hypothesis is that raw material plays a significant role driving Hi-Lo variability. Results of the
Mann-Whitney test are presented in Table 21. Mean values for raw material subgroups have been previously presented in Tables 4-6.
113
Table 21. Mann-Whitney U test results for Hi-Lo variables by raw material
Variable Description HLD - KP HLD - ON ON - KP W Ua= 2246 2243 994.5 P(2) 0.99 0.68 0.83 n= 100-45 100-43 43-45 T Ua= 2670.5 2536.5 1098.5 P(2) 0.76 0.83 0.7 n= 108-48 108-48 48-48 DEW Ua= 2353 3004 1213.5 P(2) 0.28 0.48 0.62 n= 115-46 115-56 56-46 NW Ua= 2444.5 3288 1388 P(2) 0.07. 0.15 0.65 n= 124-48 124-61 61-48 BW Ua= 2273.5 2945 1042 P(2) 0.29 0.81 0.45 n= 116-44 116-52 52-44 SW Ua= 2619 3766.5 1362 P(2) 0.93 0.82 0.52 n= 123-49 123-60 60-49 EE Ua= 2915.5 3126 1285.5 P(2) 0.23 1 0.39 n= 118-44 118-53 53-44 BLW Ua= 2915 3898.5 1372 P(2) 0.71 0.87 0.58 n= 126-48 126-61 61-48 BC Ua= 3523 4455.5 1479 P(2) 0.03 0.05 0.78 n= 124-47 124-61 61-47 TD Ua= 2562.5 3155.5 1265 P(2) 0.54 0.79 0.56 n= 112-43 112-55 55-43 BCon Ua= 3149 3728.5 1569 P(2) 0.91 0.51 0.65 n= 130-49 130-61 61-49 BCon.NW Ua= 3184.5 4001 1483 P(2) 0.48 0.52 0.91 n= 124-48 124-61 61-48 BCon.BW Ua= 2603 3092 1116 P(2) 0.85 0.65 0.97 n= 116-44 116-51 51-44 N.BW Ua= 2502.5 3095 1105.5
114
P(2) 0.79 0.86 0.78 n= 117-44 117-52 52-44 BLL Ua= 1936 2763.5 982 P(2) 0.15 0.87 0.25 n= 106-43 106-53 53-43 NH Ua= 2627 2698.5 1480 P(2) 0.49 0.1 0.05 n= 114-43 114-56 56-43 HL Ua= 2652 2790 1374 P(2) 0.7 0.47 0.36 n= 111-46 111-54 54-46 S.CL Ua= 2546 2881 1188.5 P(2) 0.75 0.91 0.75 n= 112-44 112-52 52-44 ML Ua= 2130 2897 1087 P(2) 0.21 0.42 0.5 n= 114-43 114-55 55-43 BLWH Ua= 2662 3521.5 1409 P(2) 0.22 0.35 0.74 n= 126-48 126-61 61-48 BIR Ua= 2581 3148 1274 P(2) 0.61 0.88 0.62 n= 114-43 114-56 56-43 CR Ua= 2152 2912 1302 P(2) 0.02 0.04 0.51 n= 119-47 119-60 60-47 MBA Ua= 2252 3875.5 1181 P(2) 0.01 0.5029 0.04 n= 133-49 113-62 62-49 L/W Ua= 2182 2882 1141 P(2) 0.41 0.56 0.77 n= 111-43 111-55 55-43
Results of the Mann-Whitney test indicate that few statistically significant differences exist between Hi-Lo points made from different raw materials. Statistically significant differences between Hi-Lo points principally revolve around differences between Kettle Point chert specimens and those made from Haldimand, and to a lesser extent Onondaga chert.
115
Differences in raw material workability may account for the differences between Haldimand and
Kettle Point chert Hi-Lo points as expressed by the measures of basal constriction (BC), the constriction ratio (CR) and the mean basal angle (MBA) (Table 21). Similarly, Haldimand chert specimens have also been observed to be statistically different from Onondaga chert Hi-Lo points with regards to their basal constriction and constriction ratios (Table 21). These variables will be discussed further below.
Previously a strong positive correlation has been noted between the measure of basal constriction and the constriction ratio (Table 18), as such, these variables do not individually warrant in-depth discussion. The basal constriction variable quantifies the difference in width between the shoulders and the point of minimum basal width in terms of millimeters while the constriction ratio provides a means of comparing the relative constriction between points without the confounding effects of size. That these variables vary significantly between Haldimand chert Hi-
Lo points and those made from Onondaga and Kettle Point indicates that Haldimand chert points have the greatest degree of constriction in terms of both absolute distance and relative constriction
(see Tables 15-17).
Of particular interest is the observed differences between values for the mean basal angle of
Haldimand chert specimens and Kettle Point and Onondaga chert Hi-Lo points. Kettle Point chert
Hi-Lo points have the lowest observed basal angles (mean 78.35°) when compared to those of
Onondaga chert (mean 81.47°) and Haldimand chert (mean 82.33°). This evidence indicates that
Hi-Lo points manufactured from Kettle Point chert have ears which on average expand at a somewhat more abruptly than those of Haldimand or Onondaga chert. This finding may have some significance in terms of Ellis’ (2004a) Hi-Lo subtypes. Of the three subtypes proposed, those seen as more definitively side notched are considered to be the latest manifestation of Hi-Lo and most closely related to later early Archaic forms. However, it is important to view basal angles in terms of their relation to other relevant variables such as the linear measurements of ear
116 expansion (EE) and basal constriction (BC). Lateral constriction on Hi-Lo bases exists on a continuum between weakly indented and those which would fit colloquial definitions of side notched (see Ellis and Deller 1982). Without a clear definition for side notching, as opposed to a more minor lateral indentation, clear typological distinctions can be difficult to draw for seemingly transitional forms like Hi-Lo which may represent the incipience of that technological behavior. While such a definition may seem semantic it may be necessary to further refine a type which has been elsewhere referred to as “loosely defined” (Mason 1981: 114) or a “grab bag”
(Ellis 2004a: 62). If differences in the configuration of Hi-Lo bases have temporal significance, further refinement of the Hi-Lo type may be necessary to track technological changes.
5.7 External Coefficient of Variation Comparison Many of the coefficients of variation presented above are comparable to those presented by White
(2013: Table 2) and Ellis and Deller (2014: Table 4) for identical variables (e.g. BLW and
Maximum Fore-Section Width). Each of these samples contain a significant amount of common material drawn from Middlesex County, Ontario (see Ellis and Deller 2014: Figure 1 and White
2013: Figure 5). Coefficients of variation for comparable variables employed by White (2013),
Ellis and Deller (2014) and this study are presented below in Table 22, comparisons are represented graphically by Figure 30.
Table 22. Coefficient of variation comparison for Hi-Lo
Variable White 2013 Browne 2015 Ellis and Deller 2014 DEW 20.80 19.89 - NW 16.10 15.27 - BW 15.10 14.86 - EE 64.30 74.98 - BCon 41.60 37.66 32.5 T 14.80 15.20 - NW 16.10 15.27 - SW - 13.34 10.79 ML - 21.18 18.31 BLL - 26.82 24.26 BLW - 13.72 10.76 HL - 22.03 16.07
117
80
70
60
50
40
30
20
10
0 DEW NW BW EE Bcon T NW SW ML BLL BLW HL
WHITE 2013 BROWNE 2015 ELLIS AND DELLER 2014
Figure 30. Coefficient of variation for equivalent measurement by Hi-Lo study
Different CV values for common variables should represent the variation introduced by those points not common to the three samples. This comparison offers an opportunity to reflect on the influence of typological constructs on the composition of projectile point samples.
It is worth considering that differential reference material, whether site collections or published type descriptions and examples, will play some role in determining the range of variation that one considers appropriate within the Hi-Lo type. The preferential use of a single reference for Hi-Lo definition and examples for comparison from among the works of Fitting (1963), Justice (1987),
Ellis (2004a), Ellis and Deller (1982, 2014), Ellis, Timmins and Martelle (Ellis et al. 2009) or
Payne (1982) could lead to a range of conceptions of Hi-Lo between researchers. With time and the growing size of Hi-Lo sample from new collections and research, the definition of Hi-Lo has evolved in such a way that the initial Hi-Lo type specimens of Fitting (1963) now seem to encompass a limited range of the variability considered appropriate within the modern Hi-Lo type definition. This shift is inevitable given the extensional definition of the Hi-Lo type (O’Brien and
Lyman 2002). Given that Hi-Lo points are most often reported as isolated surface finds, one’s conception of the range of variability allowed within the Hi-Lo type is particularly influential for classification. For example, it is possible Ellis’ (2004a) conception of an explicitly side-notched
118
Hi-Lo subtype may have increased the rate of definition for points with more dramatic lateral indentations. The term ‘side-notched’ itself caries a significant amount of implicit meaning unique to each researcher, which is bound to play some role in typological classifications. Could this problem play a role in driving the high CV for EE for the 2015 Hi-Lo sample which is composed largely of surface finds reported by a wide range Ontario archaeologists? Ideas such as these are difficult to evaluate, especially when CV values appear largely comparable between samples.
The 2015 Hi-Lo sample was drawn almost largely from Ontario while White’s (2013) sample was drawn primarily from the United States. Differences between commonly cited typological definitions of Hi-Lo on either side of border (see Overstreet 1993) could potentially play some role in the reporting of archaeological materials and therefore the materials available for inclusion in either study (see Section 5.2.2 for a potential example). Common typological definitions remain essential for the creation of datasets with interpretive value. The use of common definitions are vital for archaeological complexes which manifest over large geographic areas and/or across modern international borders. This problem is especially important for archaeological materials like Hi-Lo projectile points which are often recovered as isolated surface finds by a diverse range of researchers.
5.8 Principal Components Analysis
The purpose of principal component analysis (PCA) is to find the best low-dimensional representation of the variation in a multivariate data set where variables have potential for covariation. PCA involves converting observed variables into standardized, scaled-value variables which are reorganized into new orthogonal, linearly uncorrelated variables called principal components. PCA acts to sort co-varying variables (i.e. point weight and length) into principal components which are able to express the chief sources of variation in the dataset along a spectrum. Principal components are ordered so that the first principal component explains the
119 greatest amount of variation with subsequent components explaining increasingly small proportions of uncorrelated variation, i.e. PC1 > PC2 > ... PCn. Since principal components are orthogonal, any pair of principal components will have zero correlation. If the total variance represented by a principal component is 20.0% that means that PCn best “explains” the variability of 20.0% of the recorded objects from the initial data set. Thus, PCA may be used to determine the primary discriminators accounting for the greatest proportion of variation in a dataset. PCA may reveal groupings within data and be used to test the existence of hypothesized groups (i.e. raw material’s significant influence on form). In this case, the dataset was colour coded according to raw material to examine the effect of raw material on Hi-Lo point size. Separate principal components analyses were run for different sets of variables. The first PCA, the Whole Point
PCA, assesses whole point size by including all recorded variables. The second PCA analysis is focused on the base. This basal variable set, Hi-Lo Base PCA, includes only those variables which relate to aspects of base size. Separate analyses were performed to insure that basal variability would not be obscured by variability in the blade elements. Results of the principal component analyses are presented below.
120
5.9 Whole Point PCA
This section presents the results of the principal components analysis of all recorded variables.
Eigenvectors for the Whole Point PCA are presented in Table 23 while the results of the analysis are presented graphically by Figure 31 in the form of two biplots.
Table 23. Rotated eigenvector values for principal components 1, 2 and 3 in Whole Point PCA. Highlight indicates values of a magnitude greater than 0.25
Variable PC1 PC2 PC3 Description W -0.31 -0.06 0.15 T -0.21 -0.01 0.15 DEW -0.21 0.22 -0.26 NW -0.30 0.23 -0.24 BW -0.28 0.30 -0.17 SW -0.35 -0.05 -0.23 EE 0.02 0.30 0.16 BLW -0.35 -0.10 -0.20 BC -0.09 -0.36 -0.01 TD -0.05 -0.03 0.04 BCon -0.11 0.14 -0.25 BCon.N 0.02 -0.32 -0.33 BCon.BW -0.01 -0.10 -0.28 N.BW -0.03 0.34 0.22 BLL -0.27 -0.07 0.21 NH -0.05 0.11 0.07 HL -0.24 -0.04 0.08 S.CL -0.21 -0.12 0.05 ML -0.32 -0.07 0.23 BLWH -0.27 -0.10 0.20 BIR -0.06 -0.18 0.16 CR -0.01 0.40 -0.10 MBA -0.05 -0.28 -0.19 L.W -0.13 -0.03 0.39
121
Figure 31. Whole Point PCA - Principal components 1 versus 2 and principal components 1 versus 3
122
5.9.A PC1 Whole Point
Variance in the first principal component accounts for 22.9% of the variation in the dataset. PC1 can be seen as a general size component since overall size appears to be the primary discriminator between points. Variability largely relates to basic measurements of length and width for points’ features. The first component is primarily characterized by significant covariance between the neck width (NW), maximum basal width (BW), shoulder width (SW), blade width (BLW), blade length (BLL), maximum length (ML) and the height of the maximum blade width (BLWH). It is important to note that most variables, while not necessarily reaching the 0.25 level of significance, can be seen to co-vary with the main group which principally drives variability in the component.
Variability in this component is a reflection of the relationship between the principal variables listed above and those which do not co-vary significantly with this group and those variables which do not co-vary at all. It is interesting to note that the eigenvector for the depth of the basal concavity (BCon) does not reach the magnitude of significance (>0.25) in this component suggesting that larger points will not necessarily have correspondingly deep basal concavities.
This evidence suggests that as points are re-sharpened on the haft, their basal concavity will remain relatively unaltered. This result is likely an expression of the “Frison Effect” which describes how the edges of lithic tools change over the course of their use-life through resharpening (Frison 1968). Since reworking will mostly focus on the rejuvenation of working- edges it stands to reason that the basal concavities of Hi-Lo points were rarely altered after initial production. The differential effects of reworking likely account for the fact that depth of the basal concavity does not feature among those variables which are of most use in distinguishing whole
Hi-Lo points.
While almost all variables in PC1 have negative eigenvectors the distance from the point of maximum basal concavity to the neck (BCon.N) and the horizontal measure of ear expansion
123
(EE) have positive eigenvectors. Whether an eigenvector is positive or negative is not in itself significant, it does however provide a means of understanding its relationship to the other variables. Positive values for both BCon.N and EE indicate that neither of these variables is correlated with variables whose eigenvectors are negative. These values have an inverse relationship meaning that higher values in either group are associated with lower values for the others and vice versa. In this case both BCon.N and EE have only slightly inverse relationships with the main group of variables. This relationship can be interpreted as meaning that increases for the main group of variables will have little effect on the values of BCon.N or EE. Therefore for points included in PC1, general increases in size are in rare cases associated with points whose neck lies nearer to the vertical position of the point of maximum basal concavity and whose ears expand less.
Figure 32. Hi-Lo Whole Point PCA - Principal Component 1
124
5.9.B PC2 Whole Point Variance in the first principal component accounts for 16.3% of the variation in the dataset.
Variation in PC2 is more concentrated than PC1 and is focused across seven main variables. The variation is primarily characterized by an inverse relationship between two groups of variables: the first, composed of the basal constriction (BC), the distance from the point of maximum basal concavity to the neck (BCon.N) and the mean basal angle (MBA), which is inversely correlated with the second, composed of; the maximum basal width (BW), ear expansion (EE), the vertical offset of the neck from the plan of maximum basal width (N.BW), and the constriction ratio (CR).
The polarization of basal constriction (BC) and the constriction ratio (CR) is not surprising given that an inverse relationship between these two variables has previously been noted (see Table 18).
The inverse relationship between the two groups of variables in this component is of particular interest as it suggests points can be discriminated along a line of variation in hafting. Points at one end of the spectrum of variability represented by this component have a greater degree of lateral constriction (BC) between the shoulders (SW) and the minimum basal width (NW) as well as greater basal angles (MBA) which would indicates either contracting or stemmed bases. These points may be contrasted with others at the opposite end of the spectrum of variability which tend to have smaller basal widths and a greater degree of ear expansion.
125
Figure 33. Hi-Lo Whole Point PCA - Principal Component 2
5.9.C PC3 Whole Point
Variance in the third principal component accounts for 12.1% of variation in the dataset.
Variation in PC3 is more concentrated than either PC1 or PC2 with variation mainly focused across five main variables. The length/width ratio (L.W) accounts for the most variability in this component suggesting that points are primarily distinguished by the relationship between maximum length (ML) and blade width (BLW). A higher value for a point’s length/width ratio would indicate that a point is longer than it is wide. In this component, L.W has a strong inverse relationship with three vertical measurements; BCon.N, BCon.BW, BCon and DEW, a horizontal measurement. This indicates that some points in this component have blades which are much longer than they are wide and have more slight basal concavities whose vertical position closely approximates the positions of the maximum basal width (BW) and neck (NW). Conversely, points at the other end of the spectrum of variability represented by this component may be short with wide blades and deeper basal concavities whose point of maximum depth occurs relatively far from the positions of the maximum basal width (BW) and neck (NW).
126
Figure 34. Hi-Lo Whole Point PCA - Principal component 3
5.10 Hi-Lo Base PCA
This section presents the results of the principal components analysis for those variables only relating to aspects of the basal region. Eigenvectors for the Hi-Lo Base PCA are presented in
Table 24. Results of the analysis are presented in Figure 35 in the form of two biplots.
Table 24. Rotated eigenvector values for principal components 1, 2 and 3 in Hi-Lo Base PCA. Highlight indicates values of a magnitude greater than 0.25 Variable PC1 PC2 PC3 Description DEW -0.31 -0.36 0.17 NW -0.34 -0.37 0.09 BW -0.43 -0.27 0.07 EE -0.32 0.25 0.03 BC 0.24 -0.10 -0.36 BCon -0.16 -0.31 0.19 BCon.N 0.36 -0.35 0.19 BCon.BW 0.11 -0.31 0.24 N.BW -0.39 0.22 -0.08 NH -0.16 0.06 -0.16 HL -0.14 -0.23 -0.59 S.CL -0.04 -0.27 -0.56 MBA 0.28 -0.32 -0.05
127
Figure 35. Hi-Lo Base PCA - Principal component 1 versus 2 and principal component 1 versus 3
128
5.10.A PC1 Base
Variance in the first principal component accounts for 26.7% of the variation in the dataset. This component differs from the first component of the Whole Point PCA in that it cannot be considered to be a basic size component. PC1 is characterized by covariance with the width at the distal extremes of the ears (DEW), neck width (NW), basal width (BW), ear expansion (EE) and the distance between the neck and the point of maximum basal with (N.BW). These variables have a significant inverse relationship with the basal constriction (BC) and the mean basal angle
(MBA). The inverse relationships in this component, specifically as they relate to MBA, suggest that variability in this component expresses changes in the shape of the base as well as changes in linear size.
The width of the base (BW) contributes the most variability to this component. At one end of its spectrum of variability, PC1 describes bases which are wide with expanding ears. A large BW occurs far from a large NW. The segments of the lateral basal margin between the points of minimum and maximum width expand outwards from the site of the NW to form expanding ears.
The inverse relationship between EE and MBA in this component suggests that bases with a great deal of ear expansion will typically have the lesser basal angles which suggests more defined lateral indentations. The expanding ear configuration will tend to result in a high value for DEW.
On the opposite end of the spectrum lie bases with greater basal angles, a group which would include stemmed or contracting bases, which tend to have a greater degree of contraction from the shoulders (BC). For these bases, the point of maximum basal constriction (i.e. NW) lies nearer to the point of maximum basal concavity (BCon), and this correlation is shown by the high magnitude of the BCon.N eigenvector.
129
Figure 36. Hi-Lo Base PCA - Principal Component 1
5.10.B PC2 Base
Variance in the second principal component accounts for 20.1% of the variation in the dataset.
This component bears some similarity to PC1 in that the primary horizontal measurements of basal width: DEW, NW and BW all positively co-vary. While variability in this component primarily relates to the configuration of the ears it is also significantly influenced by the depth of the basal concavity (BCon), and its relation to the neck (BCon.N) and the point of maximum basal width (BCon.BW). These variables, along with the vertical distance between the shoulder and the corner (S.CL) and the mean basal angle (MBA) all positively co-vary. Variables in this component with inverse relationships to those listed above are limited to the horizontal measurement of ear expansion (EE) and the vertical distance between the neck and maximum basal width (N.BW)
On one end of its spectrum of variability, PC2 describes a group of generally wide and elongated bases with deep basal concavities that tend to have high basal angles, meaning the lateral margins are either contracting or of an approximately stemmed configuration. This group may be contrasted with another group of generally smaller bases with shallower basal concavities and
130 lower basal angles. Lower basal angles indicate that these bases will have fairly well defined lateral indentations. This configuration can be seen through the inverse relationship between EE and the positive eigenvector values for distance measurements (e.g. NW, DEW, BW, BCon and
MBA). This component suggests a group of bases for which greater lateral expansion of the ears is associated with an overall smaller base.
Figure 37. Hi-Lo Base PCA - Principal Component 2
5.10.C PC3 Base
Variance in the third principal component accounts for 14.7% of the variation in the dataset.
Variability in this component is mainly related to the length of the haft (HL) and the related shoulder to corner length (S.CL). These two variables also strongly co-vary with the horizontal measurement basal constriction (BC), which accounts for the difference in shoulder width and the neck width (NW). These variables, which have positive eigenvector values, have moderate inverse relationships with the width at the distal extremes of the ears (DEW), the depth of the basal concavity (BCon) and its distance from the location of the neck (BCon.N) and the point of maximum basal width (BCon.BW).
131
This component suggests a group whose variability exists on a spectrum between those with long, elongated hafts with relatively shallow basal concavities and bases that have shorter hafts and somewhat deeper basal concavities. Many of the points in the first group are characterized by a long, gentle lateral indentation curves originating from a shoulder which occurs at a position high on the point’s lateral margins. This conclusion is supported by the inverse relationship between
HL, S.CL, BC and the measures of BCon.N and BCon.BW. This inverse relationship suggests that these bases will tend to be more abrupt with the anatomical features of BW and NW occurring near to the shallow basal concavity (BCon). Conversely, bases with shorter hafting elements will tend to have neck widths (NW) and maximum basal widths (BW) that occur farther from the point of maximum basal concavity (BCon) than those with longer hafts.
Figure 38. Hi-Lo Base PCA - Principal Component 3
132
5.11 Spatial Patterning in Hi-Lo Variability
This section seeks to address the question of spatial patterning in Hi-Lo size. Moran’s Index (Cliff and Ord 1973; Moran 1950) was used as a measure of global spatial autocorrelation in Hi-Lo variability. Mantel correlograms were calculated to assess the impact of increasing spatial distance on the degree of autocorrelation for the variability expressed by each principal component. Spatial autocorrelation may be understood here as the concurrence of similar sorts of variability within the Hi-Lo sample in geographic space. Positive spatial autocorrelation means that similar values for variables tend to cluster in geographic space. Thus, negative spatial autocorrelation would describe a scenario where Hi-Lo points tend to be surrounded in geographic space by highly dissimilar neighbors.
Geographic patterning in Hi-Lo variability has important implications for cultural transmission during Ontario’s Late Paleoindian period. Results of these spatial analyses of Hi-Lo variability can be used to comment on the suitability of the settling in hypothesis for explaining variation in
Hi-Lo as a tool form during Ontario’s Late Paleoindian-Early Archaic transition.
5.12 Moran’s Index
Moran’s I was calculated separately for two different but similar datasets. The first dataset was composed of the first three principal components calculated from all whole point numerical variables recorded for points in the 2015 Hi-Lo sample with sufficient geographical provenience to place them at the county level (n=280). The second dataset was again composed of only points with adequate provenance in the in the 2015 Hi-Lo sample but instead included the first three principal components calculated from the set of numerical variables that only concern aspects of the hafting region (i.e. NW, BCon, MBA). These variables only relate to measurements taken from landmarks 2-10 and omit any variables which describe an aspect of the base in relation to another portion of the point (i.e. BIR).
133
This approach was taken with the understanding that Hi-Lo blade elements are highly dynamic and variable during the course of the point’s life history (Ellis and Deller 2014). This understanding informed the assumption that Hi-Lo hafting elements are less likely to have been modified after production than the blades (Fitting 1975; Ellis and Deller 1982: 2014). The hafting element is thus more likely to be encoded with stylistic elements of tool design that may have temporal significance (Ellis 2004a; White 2013). Separate assessments of spatial variability for these two datasets ensures that variability in the Hi-Lo hafting element of particular interpretative value will not be obscured by variability in the blade element.
An approach utilizing sets of PCA eigenvectors allows for object values in a multivariate dataset to be expressed as single number values in a univariate range of variability. The extracted PCA eigenvectors relating each point to a particular principal component are each values in a range of standardized values which can be used to confidently assess the similarity of points. Principal components analyses have previously partitioned the variability in Hi-Lo points into groups
(components) characterized by common relationships between variables and expressed variability within these groups on a scaled spectrum.
5.12.A Results
A positive observed value for Moran’s I indicates positive spatial autocorrelation while a negative observed value indicates negative autocorrelation. Values for Moran's I may range from −1 (an indication of a perfect dispersion of Hi-Lo variability) to +1 (perfect correlation of Hi-Lo variability in space), with a value of zero indicating wholly random spatial patterning. The values of Moran’s I for PC1-3 for the Whole Point PCA are presented below in Table 25 while results for PC1-3 for the Hi-Lo Base PCA are presented in Table 26.
134
Table 25. Moran’s I results for Whole Point PCA
PC1 PC2 PC3 Observed -0.004 -0.022 -0.0005 Expected -0.003 -0.004 -0.003 Standard 0.012 0.012 0.012 Deviation P Value 0.954 0.140 0.805
Table 26. Moran’s I results for Hi-Lo Base PCA
PC1 PC2 PC3 Observed -0.036 0.005 0.012 Expected -0.004 -0.004 -0.004 Standard 0.012 0.012 0.012 Deviation P Value 0.008 0.507 0.214
Only the P value for PC1 of the Hi-Lo base variables achieves the 0.05 level of significance indicating that for this component we can reject the null hypothesis that there is zero spatial autocorrelation. The negative value of Moran’s I for PC1 indicates a slight degree of negative spatial autocorrelation. All other components of either dataset indicate that there is insufficient reason to reject the null hypothesis. For these components Moran’s I indicates essentially random spatial patterning. Spatial patterning in PC1 of the Hi-Lo base variables and the other components is further discussed and visualized in Section 5.9.4.
5.13 Mantel Correlogram
Mantel correlograms were calculated to assess how the degree of spatial autocorrelation in the datasets described above change with increasing distance (Oden and Sokal 1986; Sokal 1986;
Bocard and Legendre 2012). The Mantel correlogram is of particular interpretive value as it expresses the effects of distance on the degree of spatial autocorrelation in the Hi-Lo sample. It provides a measure of the correlation between distance and variability within a principal component. Mantel correlograms were calculated for each dataset using two matrices of geographic distance (i.e. x, y UTM coordinates) and a numerical measure of similarity (i.e. PC1,
PC2 or PC3).
135
Similar to Moran’s I values, the Mantel correlogram expresses the degree of autocorrelation (on the y axis) as a range from −1 (an indication of a perfect negative autocorrelation) to +1 (perfect positive autocorrelation of Hi-Lo variability in space), with a zero value indicating wholly random spatial patterning (see Figure 39). Mantel correlograms are presented individually in Figures 41 and 42 respectively.
Figure 39. Mantel correlogram expressing types of spatial variability
5.13.A Results
All but one of the correlograms failed to produce clear trends of either significant positive or negative spatial autocorrelation. No components from either the Whole Point PCA or the Hi-Lo
Base PCA were found to have any measure of autocorrelation exceeding 0.04 in magnitude.
Given that these values are so close to zero they should be interpreted as exceedingly weak autocorrelation signals. Thus, spatial patterning in Hi-Lo variability is interpreted as random.
Five of the six Mantel correlograms calculated indicate no coherent trends in Hi-Lo spatial variability. The sole exception is the correlogram for PC1 of the Hi-Lo Base PCA. The trend line appears to indicate some degree of positive spatial autocorrelation for bases within 150km of each other. However, it should also be noted that this correlogram also indicates that bases <50km away from one another are less closely related than those separated by distances >50km.
Similarity appears to decline sharply for bases separated by more than 175 km. PC1 of the Hi-Lo
136 base variables is particularly useful for characterizing variability in the context of this analysis, explaining 26.7% of Hi-Lo basal variability. While this may appear promising, it should be remembered that the degree of autocorrelation being expressed is exceedingly minute (>0.02).
The interpretative value of these correlograms lie in their expression of essentially random spatial patterning in Hi-Lo projectile point variability. This interpretation can be better understood with reference to the Ontario and western New York Hi-Lo map (Figure 40), which depicts find location density through a Kernel Density Estimate (KDE) heat map with a 25 km radius kernel around finds. The isolated Kilmer site (see Tankersley et al. 1996) at the bottom right of the map provides a depiction of the 50 km diameter of these buffers.
Figure 40. Hi-Lo location heat map, (Kernel Size=1 per find location regardless of site type, 25km buffer)
Note that only the correlogram for PC1 of the Hi-Lo base variables indicates that point bases within 50 km of each other are positively autocorrelated, the other five correlograms indicate a measure of negative autocorrelation at this distance. A circular area measuring 50 km in diameter is large enough to essentially encompass any of the major clusters of finds (see Figure 40). If negative autocorrelation exists at the 50 km level it is difficult to reconcile patterning in Hi-Lo variability with the idea of geographic regionalization in point design.
137
Whole Point PC1
Whole Point PC2
Whole Point PC3
Figure 41. Mantel correlograms for PC1-3 of Whole Point PCA
138
Hi-Lo Base PC1
Hi-Lo Base PC2
Hi-Lo Base PC3
Figure 42. Mantel Correlograms for PC1-3 of Hi-Lo Base PCA
139
5.14 Regional PCA Analysis
To further test for the existence of regional difference in point size a principal components analysis (see Section 5.6) was run on both the dataset composed of the whole point variables and the dataset composed of just the basal variables. Points were sorted by location into regional groupings using the five regions specified above in Section 5.3.1 (as per Fig. 7).
5.14.A Results
Principal components 1-3 are represented graphically for each dataset by Figures 43 and 44. This analysis is concerned with detecting overall differences in variability patterning within regions rather than describing variability between regions. This result is because these regions are highly arbitrary and do not necessarily have historical meaning for Hi-Lo peoples. Thus, interpreting components beyond the scope of a basic heuristic model for characterizing overall variability between geographic regions may not be appropriate.
Principal components for both datasets do not seem to manifest in substantially different ways between regional groupings of points. It is not surprising that the ellipses for some of the less well represented regions show the least degree of similarity clustering for variation in the components.
Smaller sample sizes for regions like Niagara and western New York and North of Lake Ontario are likely most influenced by the presence of outliers. When viewed in isolation, well represented regions like Middlesex County and Southwestern Ontario, the Grand River Drainage and the Erie
Lowlands overlap substantially indicating a great degree of similarity in the spread of the variation in the components between regions. When these results are viewed in the contexts of the
Moran’s I and Mantel correlogram analyses it appears that some of the variability in the degree and nature of spatial autocorrelation across great distances may be influenced by the effects of sampling. Assessed points in the Niagara and western New York and North of Lake Ontario regions are relatively isolated and typically occur at a significant distance from major clusters of
140 points elsewhere in southern Ontario (see Fig. 14). Thus, while points may vary to some degree between regions interpretation of patterning is hindered by the effects of differential sampling.
Figure 43. Whole Point PCA - Principal components 1 versus 2 and principal components 1 versus 3 sorted by region
141
Figure 44. Hi-Lo Base PCA - Principal components 1 versus 2 and principal components 1 versus 3 sorted by region
142
5.10 Chapter Summary
The purpose of this chapter was to characterize biface variability by focusing on the possible impacts of raw material and geographic space. The main conclusions of this analysis are summarized below:
1. Statistical analyses demonstrate that there is little spatial patterning in Hi-Lo morphology
across southern Ontario and New York.
2. Tests for significant differences in size by raw material indicate that raw material
constraints are not a major determinant of morphological variability.
3. Kettle Point Hi-Lo bifaces were found to have some significant difference from
Haldimand and Onondaga chert Hi-Lo specimens in terms of ear configuration. Kettle
Point chert Hi-Lo were found to be the most definitively side notched
4. While Haldimand chert appears to be the most popular material used for points at this
time, Onondaga chert was also frequently used for Hi-Lo point production. Onondaga Hi-
Lo bifaces are more common than Kettle Point Hi-Lo specimens in each area outside of
the immediate vicinity of the Kettle Point chert outcrop.
5. Non-chert materials are in rare cases used for the production of Hi-Lo points.
6. Significant numbers of Hi-Lo points were observed in numerous local area collections in
Elgin and Norfolk Counties, Ontario suggesting that Hi-Lo points are underreported in
these areas.
7. The frequency of reported Hi-Lo points drops dramatically at the modern Ontario-New
York border.
143
Chapter 6 Conclusions
The Hi-Lo sample used for this research was assembled for the purpose of providing morphological data which could be used to interpret aspects of social organization during the late
Paleoindian period in Ontario and New York. Morphological data was derived from the Hi-Lo biface sample using landmark geometric morphometrics. Results of the morphological analysis were used to ascertain the impact of hypothesized settling in processes on Hi-Lo point manufacture. Spatial statistics were used to determine whether or not regional differences in point manufacture existed. Results of the spatial analyses were interpreted through the lens of cultural transmission theory in terms of their implications for the geographic scale of Hi-Lo social learning. The existence of clear regional differences in point manufacture was considered as a proxy for the existence of social or geographic inhibitors to cultural transmission in the study area. Hypotheses related to the geographic scale of social learning for Hi-Lo knappers were conceived of as best-fit scenarios to describe conditions potentially driving variability within a
Hi-Lo learning system:
1. Accumulated copying error and innovation in an internally unbounded learning system
2. Inherent synchronous design variation
3. Geographic regionalization of variation in an internally bounded learning system
4. Design variation driven by raw material constraints
The idea that differential raw material constraints were responsible for a significant portion of Hi-
Lo morphological variability was considered as a sort of null hypothesis for this analysis. If variability in Hi-Lo bifaces was best explained by raw material constraints, then further interpretation of cultural factors potentially driving variability would not be appropriate.
144
Morphological analyses of Hi-Lo biface size and configuration indicate that points within the type are highly variable. Variability in the sample is only weakly correlated with raw material suggesting that differential raw material constraints play a very small role driving variability within the Hi-Lo type. Kettle Point chert Hi-Lo points were observed to have the lowest observed basal angles (mean 78.35°) when compared to those of Onondaga chert (mean 81.47°) and
Haldimand chert (mean 82.33°). This indicates that Kettle Point Hi-Lo points in the sample showed the most definite ear expansion. If Ellis’ (2004a) subtypes do in fact represent a temporal sequence, this data suggests that Kettle Point chert Hi-Lo points were most commonly manufactured during the latter stages of the Hi-Lo phase. This shift may relate to the exposure of the Kettle Point chert outcrop with the draining of glacial Lake Algonquin (Janusas 1984; Karrow and Warner 1990; Fox 2009). As Kettle Point chert was an uncommon toolstone for early
Paleoindian groups, presumed ancestral to Hi-Lo groups, it is possible that the use of Kettle Point was not engrained in the cultural memory of Hi-Lo peoples. The idea of a cultural connection to early Paleoindian groups has also been evoked to explain the popularity of Haldimand chert with
Hi-Lo knappers (Fox et al. 2015). The whitish colour of Haldimand chert may have given it some level of cultural significance as its colouring is reminiscent of Collingwood chert which was central to early Paleoindian technology in Ontario. A cultural preference for white coloured cherts inherited from early Paleoindian group may have discouraged the use of Kettle Point chert by early Hi-Lo knappers. This might suggest a later date for the Welke-Tonkonoh site where Kettle
Point chert comprises a significant part of the tool assemblage (Ellis and Deller 1982: 2014).
Given that Hi-Lo is probably contemporary with northerly groups manufacturing lanceolate
Madina plano points (Fox et al. 2015), it is possible that access to the Kettle Point chert outcrop may have been restricted at times due to inter-group competition for resources. Plano groups in southern Ontario frequently made use of northern toolstone sources such as Collingwood chert
(Dibb 2004). The Heaman site, located only a short distance from the Kettle Point chert outcrop,
145 provides definitive evidence for the use of this material by plano point making peoples (Ellis and
Deller 1986: 44). It is possible that some elements of competitive exclusion of Hi-Lo groups from the Kettle Point chert outcrop may have existed during the earlier Hi-Lo phase. Exclusionary practices may have eased with time allowing Hi-Lo groups greater access to Kettle Point. This phenomenon could account for the fact that Kettle Point chert Hi-Lo points were most commonly manufactured during the latter stages of the Hi-Lo phase.
While Haldimand chert is confirmed to be the most popular material used for Hi-Lo points,
Onondaga chert was used considerably more often than previously thought (see Ellis 2004a: 59).
Onondaga Hi-Lo bifaces are more common than Kettle Point Hi-Lo in each area outside of the immediate vicinity of the Kettle Point chert outcrop. Only in the Middlesex County and
Southwestern Ontario region was Kettle Point chert used more frequently than Onondaga chert.
These findings highlight the importance of a large scale geographic perspective for the interpretation of raw material procurement patterns.
Results of the spatial analyses indicate that variability in Hi-Lo points is distributed randomly in space. Variability is best explained as the result of accumulated copying error and innovation over time in an internally unbounded learning system. These results suggest that the Hi-Lo knappers within the study area were not isolated, whether socially or geographically, from one another to an extent which would encourage the development of regional design variants. These results do not rule out the possibility that spatial patterning could exist at larger scale. Assuming that raw material was directly procured by Hi-Lo groups, chert outcrops may have provided a common venue facilitating inter-group cultural transmission. The ubiquity and continuous distribution of
Haldimand chert Hi-Lo points in Ontario suggests that this material was central to Hi-Lo point production. It is suggested that the Grand River region, home to both Haldimand and Onondaga chert outcrops may have functioned as a central place of return for highly mobile late Paleoindian groups where cultural information was shared.
146
6.1 Future Directions for Research
Further work in Elgin and Haldimand-Norfolk counties is needed to document Hi-Lo the extent of the Hi-Lo occupation in these areas. This region is critical as it situated between Kettle Point and the Haldimand and Onondaga outcrops and appears to have acted as the corridor for the movement of raw materials. A sizable Hi-Lo presence can be observed in private farm collections but the region has yet to receive serious attention from archaeologists studying the late
Paleoindian period. Further work documenting the Hi-Lo occupation of Lambton County and the regional municipality of Chatham-Kent is needed. Hi-Lo is not well represented in these areas despite the abundance of finds to the east, west and south into northwest Ohio. These areas connect the Ontario and New York sample used in this study to the sample of Hi-Lo from the
North American Midcontinent used by White (2013). This situation highlights the importance of engagement with local avocational archaeologists whose awareness of local area collections, and their provenience, may be of great value locating productive Hi-Lo sites for excavation.
The excavation of sites should be a primary goal for future Hi-Lo research. Our understanding of
Hi-Lo technology is seriously limited by the lack of dated sites in the Great Lakes. Some sites known through the literature have only been subject to partial excavation. Given the paucity of
Hi-Lo sites known, the full excavation of these sites will likely prove informative. Further excavations also have the potential to provide datable materials. Without additional temporal refinement, it is difficult to confidently place Hi-Lo technology into a paleo-environmental context which could explain aspects of late Paleoindian lifeways. The need for fine-grained paleo- environmental data is best seen at the Welke-Tonkonoh site. Environmental context is needed to begin to model the artifact discard patterns that created the Welke-Tonkonoh site. While isolated
Hi-Lo findspots may be a product of discard during the hunting and processing of terrestrial game, the unique concentration of points at Welke-Tonkonoh cannot be so easily explained. This concentration of points could be the result of group aggregation or the gradual accumulation of
147 material as Hi-Lo peoples regularly exploited a dependable resource, whether floral or faunal.
Use-wear analysis of Hi-Lo points from Welke-Tonkonoh may provide important clues for changing subsistence patterns with the onset of the Holocene and could help explain the formation processes which gave rise to the Welke-Tonkonoh site. Procrustes superimposition (see Adams et al. 2004; Buchanan and Collard 2010; Mitteroecker and Gunz 2009; Slice 2007) combined with use-wear analysis would provide a powerful analytical tool to assess functional modifications to
Hi-Lo biface design.
6.2 Research Evaluation
This study used patterning in Hi-Lo size to assess the geographic scale of social learning in the late Paleoindian period. It is important to remember that measurements of size can only approximate one half of an artifact’s form. This research does not rule out the potential that Hi-Lo variability may display spatial patterning when viewed from the perspective of shape. Thus it is possible that the linear measurements which formed the bulk of the variables in this analysis are unable to detect existing spatial patterning in Hi-Lo variability. This represents a major limitation of the study. The use of generalized procustes analysis is becoming increasingly common among archaeologists studying biface variability. This research could be made more robust through the use of linear measurements in combination with procustes analysis to capture both Hi-Lo size and shape.
This work should be considered in light of sampling issues which may have had some influence on the final result. The inability to include the points attested by Roberts (1985) and Tinkler and
Pengelly (2004) in the study sample hindered efforts to create a continuous distribution of points throughout the study area. These two sets represent a considerable number of points each from underrepresented areas. Examination of these points was needed to comment on the use of
Ancaster chert by Hi-Lo knappers.
148
In the future, this sample should be integrated with that of White (2013). It is possible that the development of regional design variants could be taking place on a larger scale than was considered in this study. A larger sample with a greater geographic distribution may provide a suitable opportunity to test macro-responses to risk and selection.
Further work is needed to place the Hi-Lo findspots documented in this study in their physiographic settings. This work may be useful for the evaluation of land use patterns. While developing an understanding of specific land-use patterns was not a stated goal of this research it formed an object of considerable interest which would certainly relate to the transmission of cultural information between Hi-Lo knappers.
6.3 Research Value
This study provides valuable information relating to:
1. The overall geographic distribution of Hi-Lo points in Ontario and New York.
2. The underreported Hi-Lo occupation in Elgin and Norfolk Counties.
3. Raw material usage for Hi-Lo point manufacture.
4. The use of geometric morphometrics as a technique to assess artifact size and
configuration.
5. The impact of raw material constraints on Hi-Lo manufacture.
6. The aspects of size and configuration that best characterize the range of variability within
the Hi-Lo type.
7. Spatial patterning in Hi-Lo point variability.
149
6.4 Final Remarks
The documentation and interpretation of variability among specimens assigned to existing typologies is an important avenue of research for archaeology in the 21st century. Typological classifications deserve continuing analysis and should not be considered as set in stone. The growth of CRM archaeology has dramatically increased the volume of archaeological research in
North America. Ultimately, the value of this work is contingent on the availability of site reports for future researchers. This thesis involved the use of a substantial number of CRM site reports for images of Hi-Lo points and their provenience. It is important for researchers to engage with
‘grey literature’ in order to fully consider the known archaeological record. Demonstrating a demand for this work and its value for research will only improve avenues of access for information in the future. Graduate students and academic archaeologists unencumbered by the rigors and time constraints of consulting archaeology must make a greater effort to utilize this information. Failure to utilize the data generated by the consulting archaeologist community does a disservice to the archaeological record and those who labored to make it known.
150
References Cited: ARA (Archaeological Research Associates Ltd.) 2004 Stage 2-3 Archaeological Assessment Seaton Lands – Block H Part Lots 17-21, Concessions 4-5 Township of Pickering, Ontario. Manuscript on file, Ontario Ministry of Tourism, Culture and Sport, Toronto. 2010 Stage 1 and 2 Archaeological Assessment North Burgess Solar Project (FIT – F0HJPWL) Township of Tay Valley, Lanark County, Ontario. Manuscript on file, Ontario Ministry of Tourism, Culture and Sport, Toronto. Adams D, F.J. Rohlf and D.E. Slice. 2004 Geometric Morphometrics: Ten Years of Progress Following the ‘Revolution’. Italian Journal of Zoology 71: 5-16. Amick, D.T. 1996 Regional Patterns of Folsom Land Use in the American Southwest. World Archaeology 27:411–426. Amick, D., T. Loebel, J.E Morrow and T.A Morrow. 1997 Hawks nest: A New Gainey-Clovis Site in Northeastern Illinois. Current Research in the Pleistocene 14:4–6. Anderson, D.G 1990 The Paleoindian Colonization of Eastern North America: a View from the Southeastern United States. In Early Paleoindian Economies of Eastern North America. Edited by K.B. Tankersley, K.B., and B.L., pp 163-216. Research in Economic Anthropology. JAI Press, Greenwich, Connecticut. 1995 Paleoindian Interaction Networks in the Eastern Woodlands. In Native American Interactions: Multiscalar Analyses and Interpretations in the Eastern Woodlands. Edited by M. S. Nassaney and K. E. Sassaman, pp. 3-26. The University of Tennessee Press, Knoxville. Anderson, D. G., and M. K. Faught 1998 The Distribution of Fluted Paleoindian Projectile Points: Update 1998.” Archaeology of Eastern North America 26: 163–187.
Anderson, D. G., and J. C. Gillam 2000 Paleoindian Colonization of the Americas: Implications from an Examination of Physiography, Demography, and Artifact Distribution. American Antiquity 65 (1): 43–66. Andrefsky, W. 2005 Lithics: Macroscopic Approaches to Analysis. 2nd ed. Cambridge University Press, New York. Bamforth, D.B. and P. Bleed 1997 Technology, Flaked Stone Technology, and Risk. In Rediscovering Darwin: Evolutionary Theory and Archeological Explanation. Edited by C. M. Barton and G. A. Clark, pp. 109- 139. Archeological Papers of the American Anthropological Association No. 7.
151
Barceló J. 2010 Visual Analysis in Archaeology. An Artificial Intelligence Approach. In Morphometrics for Nonmorphometricians. Edited by A. Elewa, pp 93 – 156, Springer, New York Barton, M.C and G.A. Clark. 1997 Stone Tools, Style, and Social Identity: an Evolutionary Perspective on the Archaeological Record. In Rediscovering Darwin: Evolutionary Theory and Archaeological Explanation. Edited by M.C. Barton and G.A. Clark., pp 141-156. Archaeological Papers of the American Anthropological Association No. 7. Bentley, R.A., and S.J. Shennan 2003 Cultural Evolution and Stochastic Network Growth. American Antiquity 68, 459–485. Bergerud, A.T. 1974 Decline of Caribou in North America Following Settlement. The Journal of Wildlife Management, 38(4):757-770 Bettinger, R.L. 2008 Cultural Transmission and Archaeology. In Cultural Transmission in Archaeology: Issues and case Studies. Edited by M.J. O’Brien, pp 10-20. Society for American Archaeology, Washington D.C. Bettinger, R.L., R. Boyd and R. Richardson 1996 Style, Function, and Cultural Evolutionary Processes. In Darwinian Archaeologies. Edited by H. D.G. Maschner, pp 133-164. New York, Plenum. Binford, L.R. 1983 Long Term Land Use Patterns: Some Implications for Archaeology. In Lulu Linear Punctated: Essays in Honor of George Irving Quimby. Edited by R.C. Dunnell and D.K Grayson, pp 27-53. Anthropological Papers. Museum of Anthropology, University of Michigan No. 72, Blackith R.E. 1957 Polymorphism in Some Australian Locusts and Grasshoppers. Biometrics 13:183–196 Blackith R.E. and Reyment R.A. 1971 Multivariate Morphometrics. Academic Press, London Bocard, D. and P. Legendre 2012 Is the Mantel Correlogram Powerful Enough to be Useful in Ecological Analysis? A simulation study. Ecology 93(6): 1473–1481 Bookstein F. L., 1989 Principal Warps: Thin-Plate Splines and the Decomposition of Deformation. IEEE Transactions on Pattern Analysis & Machine Intelligence. 11: 567–585. 1991 Morphometric tools for landmark data: geometry and biology. Cambridge University Press, Cambridge. Bookstein F.L., B. Chernoff, R.L. Elder, J.M Humphries, G.R. Smith, R.E. Strauss, 1985 Morphometrics in Evolutionary Biology. Special publication 15. Academy of Natural Sciences Press, Philadelphia
152
Boulanger, M.T. and R.L. Lyman. 2014 Northeastern North American Pleistocene Megafauna Chronologically Overlapped Minimally with Paleoindians. Quaternary Science Reviews 85:35-46 Boyd R. and P.J. Richerson 1985 Culture and the Evolutionary Process. University of Chicago Press, Chicago Boyle, D. 1906 Notes on Some Specimens. Flints. Annual Archaeological Report of Ontario 1905. Appendix to the Report of the Minister of Education, pp. 10-12. Buchanan, B. and M. Collard 2007 Investigating the Peopling of North America through Cladistic Analyses of Early Paleoindian Projectile Points. Journal of Anthropological Archaeology, 26(3):366-393. 2008 Phenetics, Cladistics, and the Search for the Alaskan Ancestors of the Paleoindians: a Reassessment of Relationships among the Clovis, Nenana, and Denali Archaeological Complexes. Journal of Archaeological Science 35:1683-1694. 2010 A Geometric Morphometrics-Based Assessment of Blade Shape Differences among Paleoindian Projectile Point Types from Western North America. Journal of Archaeological Science 37: 350–359. Buchanan, B., M.J. O'Brien and M. Collard 2014 Continent-Wide or Region-Specific? A Geometric Morphometrics-Based Assessment of Variation in Clovis Point Shape. Archaeological and Anthropological Sciences 6(2):145- 162. Bursey, J. 1998 Surface Collected Artifacts from the Murray 2 Site (AfGx-72), A Hi-Lo Component near Cayuga, Ontario. Kewa 98:2-10. Bradley, J.W., A. Speiss, R. Boisvert and J. Boudreau 2008 What’s the Point? Modal Forms and Attributes of Paleoindian Bifaces in the New England-Maritimes Region. Archaeology of Eastern North America 36:119-172. Bradley, B. A., M. B. Collins, and A. C. Hemmings 2010 Clovis Technology. Ann Arbor, MI: International Monographs in Prehistory, Archaeological Series 17. Cannon M.D. and D.J. Meltzer. 2008 Early Paleoindian Foraging: Examining the Faunal Evidence for Large Mammal Specialization and Regional Variability in Prey Choice. Quaternary Science Reviews 23:1955–1987. Cardillo M. 2010 Some Applications of Geometric Morphometrics to Archaeology. In Morphometrics for Nonmorphometricians. Edited by A. Elewa, pp 325 – 344, Springer, New York Cavalli-Sforza L.L. and M.W. Feldman 1981 Cultural Transmission and Evolution: a Quantitative Approach. Princeton University Press, Princeton.
153
Chapman, L.J. and D.F. Putnam 1984 The Physiography of Southern Ontario, 3rd Edition. Toronto: Ontario Geological Survey, Special Volume 2. Clarke, D. L. 1968 Analytical Archaeology. Methuen and Co., London, UK. Cliff, A.D. and J. K. Ord 1973 Spatial Autocorrelation. Pion, London Collard, M., M. Kenery and S. Banks 2005 Causes of Toolkit Variation among Hunter-Gatherers: A Test of Four Competing Hypotheses. Canadian Journal of Archaeology 29: 1–19 Collard, M., S.J. Shennan, B. Buchanan and R.A. Bentley 2008 Evolutionary Biological Methods and Cultural Data. In Handbook of Archaeological Theories. Edited by R.A. Bentley and H. Maschner, pp 203 - 223. AltaMira Press, Lanham. Collard, M., B. Buchanan, J. Morin and A. Costopolous 2011 What Drives the Evolution of Hunter–Gatherer Subsistence Technology? A Reanalysis of the Risk Hypothesis with Data from the Pacific Northwest. Phil. Trans. R. Soc. B (2011) 366, 1129–1138 Conolly, J 2015 Regional variability in Laurentian Archaic Biface Forms (Draft MS). Corti, M. 1993 Geometric morphometrics: An Extension to the Revolution. Trends in Ecology and Evolution 8(2): 302–303. Curran, M.L. and J.R. Grimes. 1989 Ecological Implications for Paleoindian Lithic Procurement Economy in New England. In Eastern Paleoindian Lithic Resource Use. Edited by C.J.Ellis and J.L. Lothrop, J.L., pp 41-74. Westview Press, Boulder, Colorado. Darwin, C. 1859 The Origin of Species. London, Penguin. Deller, D. B. 1976 Paleo-Indian Locations on Late Pleistocene Shorelines, Middlesex County, Ontario. Ontario Archaeology 26: 3-19 1979 Paleo-Indian Reconnaissance in the Counties of Lambton and Middlesex, Ontario. Ontario Archaeology 32:3-20. 1988 The Paleo-Indian Occupation of Southwestern Ontario: Distribution, Technology and Social Organization. Unpublished PhD Dissertation, Dept. of Anthropology, McGill University, Montreal, Quebec. 1989 Interpretation of Chert Type Variation in Paleoindian Industries, Southwestern Ontario. In Eastern Paleoindian Lithic Resource Use. Edited by C.J. Ellis and J.C. Lothrop, pp 191- 220. Westview Press, Boulder, Colorado, pp. 191–220.
154
Deller, D.B. and Ellis, C.J., 1984 Crowfield: A Preliminary Report on a Probable Paleo- Indian Cremation in Southwestern Ontario. Archaeology of Eastern North America 12:41-71. 1988 Early Paleoindian Complexes in Southwestern Ontario. In Late Pleistocene and Early Holocene Paleoecology and Archaeology of the Eastern Great Lakes Region. Edited by R. S. Laub, N. G. Miller and D. W. Steadman, pp 251-263. Bulletin of the Buffalo Society of Natural Sciences 33. 1992a Thedford II: a Paleo-Indian Site in the Ausable River Watershed of Southwestern Ontario. Memoirs of the Museum of Anthropology. University of Michigan No. 24, Ann Arbor, 157 pp. 1992b The Early Paleo-Indian Parkhill phase in Southwestern Ontario. Man in the Northeast 44:15-54. 1996 The Bolton Site (AfHj-89): A Crowfield Phase Early Paleo-Indian Site in Southwestern Ontario. Ontario Archaeology 61:5-40. 2011 The Crowfield Site (AfHj-31): A Unique Paleoindian Fluted Point Site in Southwestern Ontario. Memoirs Series, Museum of Anthropology, University of Michigan No. 49. Ann Arbor, Michigan Dibb, G. C. 1985 Late Palaeo-Indian Settlement Patterns along the Margins of the Simcoe Lowlands in Southcentral Ontario. MA Thesis, Department of Anthropology, Trent University, Peterborough, Ontario. 2004 The Madina Phase: Late Pleistocene-Early Holocene occupation along the Margins of the Simcoe Lowlands in South-Central Ontario. In The Late Palaeo-Indian Great Lakes: Geological and Archaeological Investigations of Late Pleistocene and Early Holocene Environments. Edited by Lawrence J. Jackson and Andrew Hinshelwood, pp 117 - 162. Canadian Museum of Civilization, Gatineau, Québec. Mercury Series, Paper No. 165. Dibb G.C. and C.J. Ellis 1988 Madina Plano Points. Kewa 88(7) Dickson, P.S. 2011 Hi-Lo Lithic Toolkits: New Insights from the Double Take Site. Ontario Archaeology 91:32-57 Dunnell, R.C. 1978 Style and Function: A Fundamental Dichotomy. American Antiquity 43:192-202. Eerkens, J.W., R.L. Bettinger and P.J. Richerson 2014 Cultural Transmission Theory and Hunter-Gather Archaeology. In The Oxford Handbook of the Archaeology of Anthropology of Hunter-Gatherers. Edited by V. Cummings, P. Jordan and M. Zvelebill, pp1127-1142. Oxford University Press, Oxford Eerkens J.W. and C.P. Lipo 2005 Cultural Transmission, Copying Errors, and the Generation of Variation in Material Culture and the Archaeological Record. Journal of Anthropological Archaeology 24:316– 334 2007 Cultural Transmission Theory and the Archaeological Record: Providing Context to Understanding Variation and Temporal Changes in Material Culture. Journal of Archaeological Research 15:239–274
155
Eley, B.E. and P.H. von Bitter 1989 Cherts of Southern Ontario. Royal Ontario Museum Publications, Toronto, Ontario Ellis, C.J. 1981a Hi-Lo Points. Kewa 81(2) 1981b Investigations of a Transitional Paleo-Indian to Early Archaic Lithic Industry in Southwestern Ontario. Report on file, Ontario Ministry of Culture, Heritage Branch, Toronto 1984a Gainey Fluted Points. Kewa 84(7) 1984b Barnes Fluted Points. Kewa 84(6) 1984c Crowfield Fluted Points. Kewa 84(5.5) 1984d Paleo-Indian Lithic Technological Structure and Organization in the Lower Great Lakes Area: A First Approximation. Unpublished PhD Dissertation, Department of Archaeology, Simon Fraser University, Burnaby, British Columbia, Canada. 1986 Holcombe Points. Kewa 86(7) 1989 The Explanation of Northeastern Paleoindian Lithic Procurement Patterns. In Eastern Paleoindian Lithic Resource Use. Edited by C.J. Ellis and J.C. Lothrop, pp 139-164. Westview Press, Boulder, Colorado. 2004a Hi-Lo – An Early Lithic Complex in the Great Lakes Region. In The Late Palaeo-Indian Great Lakes: Geological and Archaeological Investigations of Late Pleistocene and Early Holocene Environments. Edited by Lawrence J. Jackson and Andrew Hinshelwood, pp 57 - 84. Mercury Series Paper 165, Canadian Museum of Civilization, Ottawa 2004b The Hi-Lo Component at the Welke-Tonkonoh Site, Area C. Kewa, 5:1-19 2009 The Crowfield and Caradoc Sites, Ontario: Glimpses of Palaeo-Indian Sacred Ritual and World View. In Painting the Past with a Broad Brush: Papers in Honour of James Valliere Wright. Edited by David L. Keenlyside and Jean-Luc Pilon, pp 319-349. Mercury Series Archaeology Paper 170. Canadian Museum of Civilization, Gatineau. 2011 Measuring Paleoindian range mobility and Land-Use in the Great Lakes/Northeast. Journal of Anthropological Archaeology 30:385–401 Ellis, C.J., D.H. Carr and T.J. Loebel 2011 The Younger Dryas and Late Pleistocene peoples of the Great Lakes region. Quaternary International 242:534-545 Ellis, C. J., and D. B. Deller 1982 Hi-Lo Materials from Southwestern Ontario. Ontario Archaeology 38:3-22. 1986 Post-Glacial Lake Nipissing "Waterworn" Assemblages from the Southeastern Huron Basin Area. Ontario Archaeology 45:39-60. 1988 Some Distinctive Paleo-Indian Tool Types from the Lower Great Lakes Region. Midcontinental Journal of Archaeology 13:111-158. 1990 Paleo-Indians. In The Archaeology of Southern Ontario to A. D. 1650. Edited by C. J. Ellis and Neal Ferris, pp. 37-63. Occasional Publication 5. London Chapter, Ontario Archaeological Society. 1996 William Burton Roosa 1923-1994. Canadian Journal of Archaeology 20:135-137 1997 Variability in the Archaeological Record of Northeastern Early Paleo-Indians: a View from Southern Ontario. Archaeology of Eastern North America 25, 1–30. 2000 An Early Paleo-Indian Site near Parkhill, Ontario. Canadian Museum of Civilization, Mercury Series No. 159, Gatineau, Quebec
156
2002 Excavations at the Caradoc site (AfHj-104): A Late Paleo-Indian Ritual Artifact Deposit. Occasional Publication 8, London Chapter, Ontario Archaeological Society. 2014 Hi-Lo Point Life Histories. Kewa 4:1-39 Ellis, C.J., I.T. Kenyon and M.W. Spence 1990 The Archaic. In The Archaeology of Southern Ontario to A.D. 1650. Edited by C.J. Ellis and N. Ferris, pp 65-124. Occasional Publication 5. London Chapter, Ontario Archaeological Society Ellis, C.J., P.A. Timmins and H. Martelle 2009 At the Crossroads and Periphery: The Archaic Archaeological Record of Southern Ontario. In Archaic Societies: Diversity and Complexity Across the Midcontinent. Edited by T.E. Emerson, A. Fortier and D. McElrath, pp 787–840. State University of New York Press, Albany. Faught, M.K. 2008 Archaeological Roots of Human Diversity in the New World: A Compilation of Accurate and Precise Radiocarbon Ages from Earliest Sites. American Antiquity 73(4): 670-698 Ferring C.R. 2001 The Archaeology and Paleoecology of the Aubrey Clovis site (41DN479) Denton County, Texas. Centre for Environmental Archaeology, Department of Geography, University of North Texas Fiedel, S. 1999 Older Than We Thought: Implications of Corrected Dates for Paleoindians. American Antiquity 64(1): 95-115 Figgins, J. 1927 The Antiquity of Man in North America. Natural History 27: 229–239. 1933 A Further Contribution to the Antiquity of Man in America. Proceedings of the Colorado Museum of Natural History, 12(2), Denver, CO. Fitting, J. E. 1963 The Hi-Lo Site: A Late Paleo-Indian Site in Michigan, Wisconsin Archaeologist 44(2):87- 96. 1968 Environmental Potential and the Postglacial Readaptation in Eastern North America. American Antiquity 33(4): 441-445 1970 The Archaeology of Michigan. Natural History Press, New York. 1975 The Archaeology of Michigan: A Guide to the Prehistory of the Great Lakes Region (revised edition). Bloomfield Hills, MI., Cranbrook Institute of Science. Fitting, J.E., J. DeVisscher and E.S. Wahla 1966 The Paleo-Indian Occupation of the Holcombe Beach. Anthropological Papers, Museum of Anthropology, University of Michigan, No. 27. Ann Arbor. Fitzhugh, B. 2000 Risk and Invention in Human Technological Evolution. Journal of Anthropological Archaeology 20:125–167
157
Fox, W.A. 1979 Southern Ontario Chert Sources. Paper presented at the 11th annual meeting of the Canadian Archaeological Association, Québec City, Quebec. 1989 Book Review of “Cherts of Southern Ontario.” Arch Notes 89–6:24–27. 2009 Ontario Cherts Revisited. In Painting the Past with a Broad Brush: Papers in Honour of James Valliere Wright. Edited by David L. Keenlyside and Jean-Luc Pilon, pp 353-369. Mercury Series Archaeology Paper 170. Canadian Museum of Civilization, Gatineau. Fox, W.A. and D. B. Deller, C.J. Ellis 2015 Chert Sources and Utilization in the Southern Huron Basin during the Early Holocene. In Caribou Hunting in the Upper Great Lakes: Archaeological, Ethno- graphic, and Paleoenvironmental Perspectives. Edited by Elizabeth Sonnenburg, Ashley K. Lemke, and John M. O’Shea, pp.67-72. Memoir No. 57. Museum of Anthropological Archaeology, University of Michigan, Ann Arbor, Michigan. Frison, G.C. 1976 The Chronology of Paleoindian and Altithermal Cultures in the Big Horn Basin, Wyoming. In Cultural Change and Continuity, Essays in Honor of James Bennett Griffin. Edited by C. E. Cleland, pp. 147–173. Academic, New York. 1989 Experimental Use of Clovis Weaponry and Tools on African Elephants. American Antiquity 54: 766–784 Frison, G.C. and D.J. Stanford D J, (eds) 1982 The Agate Basin Site. Academic, New York. Garrad, C. 1971 Ontario Fluted Point Survey. Ontario Archaeology 16:3-18. Gillespie, J.D. 2007. Enculturing an Unknown World: Caches and Clovis Landscape Ideology. Canadian Journal of Archaeology 31: 171–189. Gingerich J.A, S.B. Sholts, S. Wärmländer and D. Stanford 2014 Fluted Point Manufacture in Eastern North America: an Assessment of Form and Technology Using Traditional Metrics and 3D Digital Morphometrics. World Archaeology 46(1): 101–122 Goebel, T., M.R. Waters and D.H. O'Rourke 2008 The Late Pleistocene Dispersal of Modern Humans in the Americas. Science 319:1497- 1502 Goodyear, A.C. 1982 The Chronological Position of the Dalton Horizon in the Southeastern United States. American Antiquity 47:382-395. 1989 A Hypothesis for the Use of Cryptocrystalline Raw Materials among Paleo-Indian Groups of North America. In Eastern Paleoindian Lithic Resource Use, edited by C.J. Ellis and J.C. Lothrop Westview Press, pp. 1-10 Boulder, Colorado. 2006 Recognizing the Redstone Fluted Point in the South Carolina Paleoindian Point Database. Current Research in the Pleistocene 23: 112–114.
158
Grimes, J.R. 1979 A New look at Bull Brook. Anthropology 3:109-130 Hamilton, M.J. and B. Buchanan 2007 Spatial Gradients in Clovis-age Radiocarbon Dates across North America Suggest Rapid Colonization from the North. PNAS 104:15629–15634. 2009 The Accumulation of Stochastic Copying Errors Causes Drift in Culturally Transmitted Technologies: Quantifying Clovis Evolutionary Dynamics. Journal of Anthropological Archaeology 28:55–69 Hayden, B., N. Franco and J. Spafford 1996 Evaluating Lithic Strategies and Design Criteria. In Stone Tools: Theoretical Insights into Human Prehistory. Edited by G.H. Odell, pp. 9-50. Plenum Press, New York. Haynes, G., G. Anderson, C.R. Ferring, S.J. Feidel, D.K. Grayson, C.V. Haynes, V.T. Holliday, B.B. Huckell, M. Kornfeld, D.J. Meltzer, J. Morrow, T. Surovell, N.M. Waguespack, P. Wigand and R.M Yohe 2007 Comment on “Redefining the Age of Clovis: Implications for the Peopling of the Americas”. Science 317:320b Haynes, C.V. 1964 Fluted Projectile Points: their Age and Dispersion. Science 145:1408–1413 1980 The Clovis Culture. Canadian Journal of Anthropology 1(1): 115-21 1982 Were Clovis progenitors in Beringia? In Paleoecology of Beringia. Edited by D.M. Hopkins, J. Matthews, C. Schweger and S. Young (Eds.), pp 383-398 Academic Press, New York. Haynes, C.V., D.J. Donahue, A. Zull and T. Zabel 1984 Application of Accelerator Dating to Fluted Point Sites. Archaeology of Eastern North America, 12:184-191 Henrich, J. 2004 Demography and Cultural Evolution: How Adaptive Cultural Processes can Produce Maladaptive Losses: The Tasmanian Case. American Antiquity, 69(2):197-214 Holliday, V.T. 2000 The Evolution of Paleoindian Geochronology and Typology on the Great Plains. Geoarchaeology 15(3):227–290. Holliday, V.T., E. Johnson, H. Haas, and R. Stuckenrath. 1983 Radiocarbon Ages from the Lubbock Lake Site, 1950–1980: Framework for Cultural and Ecological Change on the Southern High Plains. Plains Anthropologist 28:165–182. Houston, M and D. Perelli 2009 Broken Clock Site (UB 2488) Data Recovery Project Town of Cheektowaga, Erie County, New York. New York State Archaeological Association Newsletter 5(1):1-3 Howard, C.D. 1990 The Clovis Point: Characteristics and Type Description. Plains Anthropologist 35,129:255-262
159
Jackson, L. J. 1983 Geochronology and Settlement Disposition in the Early Palaeo-Indian Occupation of Southern Ontario, Canada. Quaternary Research 19:388-399. 1998 The Sandy Ridge and Halstead Paleo-Indian Sites, Museum of Anthropology Memoirs No. 32, University of Michigan, Ann Arbor. 2004 Changing our Views of the Late Palaeo-Indian in Ontario. In The Late Palaeo-Indian Great Lakes: Geological and Archaeological Investigations of Late Pleistocene and Early Holocene Environments. Edited by L. J. Jackson and Andrew Hinshelwood. Mercury Series Archaeological Paper 165, Canadian Museum of Civilization Jackson, L.J., C.J. Ellis, A.V. Morgan and J.H. McAndrews 2000 Glacial Lake Levels and Eastern Great Lakes Palaeo-Indians. Geoarchaeology 15:5, 415– 440 Jackson, L.J., H. McKillop, and S. Wurtzburg 1987 The Forgotten Beginning of Canadian Palaeo-Indian Studies, 1933 – 1935. Ontario Archaeology 47: 5-18 Janusas, S. E. 1984 A Petrological Analysis of Kettle Point Chert and Its Spatial and Temporal Distribution in Regional Prehistory. Mercury Series Archaeology Paper 128. National Museum of Man, Ottawa. Jennings, T.A. 2010 Exploring the San Patrice Lanceolate to Notched Hafting Transition. In Exploring Variability in Early Holocene Hunter-Gatherer Lifeways. Edited by S. Hurst and J. L. Hofman, pp. 153-166. University of Kansas Publications in Anthropology 25. Jordan P, and S.J. Shennan 2003 Cultural Transmission, Language and Basketry Traditions amongst the California Indians. Journal of Anthropological Archaeology 22:42–74. Julig, P.J. and G. Beaton 2015 Archaeology of the Late-Paleoindian/Early Archaic in the Lake Huron Region, with New Data from the Sheguiandah Site in Caribou Hunting in the Upper Great Lakes: Archaeological, Ethnographic, and Paleoenvironmental Perspectives. Edited by E. Sonnenburg, A.K. Lemke and J.M O’Shea, pp 53-66. Museum of Anthropology Memoirs No. 57, University of Michigan, Ann Arbor. Justice, N.D. 1987 Stone Age Spear and Arrow Points of the Midcontinental and Eastern United States: A Modern Survey and Reference. Indiana University Press, Bloomington. Karrow, P.F. and B.G. Warner 1990 The Geological and Biological Environment for Human Occupation in Southern Ontario. In The Archaeology of Southern Ontario to A. D. 1650. Edited by C. J. Ellis and Neal Ferris, pp. 5-36. Occasional Publication 5. London Chapter, Ontario Archaeological Society.
160
Karrow, P.F., T.W. Anderson, A. Clarke, L.D. Delorme and M. Sreenivasa 1975 Stratigraphy, Paleontology and Age of Lake Algonquin Deposits in Southwestern Ontario, Canada. Quaternary Research 5: 49-87. Kelly, R. L. 1995 The Foraging Spectrum: Diversity in Hunter–Gatherer Lifeways. Smithsonian Institution Press, Washington, DC. Kelly, R.L. and L.C. Todd 1988 Coming into the Country: Early Paleoindian Hunting and Mobility. American Antiquity 53(2):231-244 Koldehoff, B. and T. J. Loebel. 2009 Clovis and Dalton: Unbounded and Bounded Systems in the Midcontinent of North America. In Lithic Materials and Paleolithic Societies. Edited by B. Adams and B.S. Blades, pp. 270-287. Wiley-Blackwell, Chichester, UK Koldehoff, B. and J. A. Walthall. 2004 Settling In: Hunter-Gatherer Mobility during the Pleistocene-Holocene Transition in the Central Mississippi Valley. In Aboriginal Ritual and Economy in the Eastern Woodlands: Essays in Honor of Howard D. Winters. Edited by A. Cantwell, L.A. Conrad, and J.E. Reyman, pp. 49-72. Illinois State Museum Scientific Papers, Springfield, Illinois 2009 Dalton and the Early Holocene Midcontinent: Setting the Stage. In Archaic Societies: Diversity and Complexity Across the Midcontinent. Edited by T.E. Emerson, A. Fortier and D. McElrath, pp 137-151. State University of New York Press, Albany. Kidd, K.E. 1951 Fluted Points in Ontario. American Antiquity 16:260. Laird, H. C. 1935 The Nature and Origin of Chert in the Lockport and Onondaga Formations of Ontario (with Plates XVI to XXII). Transactions of the Royal Canadian Institute XX(II): 231– 304. Long, D. 2004 A Workability Comparison of Three Ontario Cherts (A Flint Knapper’s Perspective). Kewa 04-7 & 04-8:20-23 Lothrop, J.C. and J.W. Bradley 2012 Paleoindian Occupations in the Hudson Valley, New York. In Late Pleistocene Archaeology and Ecology in the Far Northeast. Edited by Claude Chapdelaine and R. A. Boisvert, pp. 9-47. Texas A&M University Press, College Station, Texas. Lycett S.J. 2009 Quantifying Transitions: Morphometric Approaches to Palaeolithic Variability and Technological Change. In Sourcebook of Paleolithic Transitions. Edited by M. Camps and P. Chauhan, pp 79-92, Springer Science. Macdonald, G.F. 1968 Debert: A Palaeo-Indian Site in Central Nova Scotia. National Museums of Canada, Anthropological Papers No. 16.
161
Marcus L. F., 1990 Traditional Morphometrics. In Proceedings of the Michigan Morphometrics Workshop. Edited by F. J. Rohlf & F. L. Bookstein, pp. 77-122. Spec. Publ. No. 2. University of Michigan Museum of Zoology, Ann Arbor. Mason, R. J. 1962 The Paleo-Indian Tradition in Eastern North America. Current Anthropology 3: 227–228. 1981 Great Lakes Archaeology. New York: Academic Press. McAndrews, J.H. 1981 Late Quaternary Climate of Ontario: Temperature Trends from the Fossil Pollen Record. In Quaternary Paleoclimate. Edited by W. C. Mahaney, pp. 319-333. Geo Abstracts, Norwich, England. 1994 Pollen Diagrams for Southern Ontario Applied to Archaeology. In Great Lakes Archaeology and Paleoecology: Exploring Interdisciplinary Initiative for the Nineties. Edited by R. MacDonald, pp 179-196. Quaternary Sciences Institute, University of Waterloo. McCarthy F. M., J. H. McAndrew and E. Papangelakis 2015 Paleoenvironmental Context for Early Holocene Caribou Migration on the Alpena- Amberey Ridge. In Caribou Hunting in the Upper Great Lakes: Archaeological, Ethnographic, and Paleoenvironmental Perspectives. Edited by E. Sonnenburg, A. Lemke and J. O’Shea, pp 13-29. Memoir No. 57. Museum of Anthropological Archaeology, University of Michigan, Ann Arbor, Michigan. Meltzer, D.J. 1984 On stone Procurement and Settlement Mobility in Eastern Fluted Point Groups. North American Archaeologist 6:1‐24 1988 Late Pleistocene Human Adaptations in Eastern North America. Journal of World Prehistory 2:1–52. 1989 Was Stone Exchanged among Northeastern Paleo-Indians? In Eastern Paleoindian Lithic Resource Use. Edited by C.J. Ellis and J.C. Lothrop, pp 11-40. Westview Press, Boulder, Colorado. 2009 First Peoples in a New World: Colonizing Ice Age America. University of California Press, Berkeley Mesoudi, A. 2011 Cultural Evolution: How Darwinian Theory can explain Human Culture and Synthesize the Social Sciences. University of Chicago Press. Chicago.
Mesoudi, A and M.J. O’Brien 2008 The Cultural Transmission of Great Basin Projectile-Point Technology: An Experimental Simulation. American Antiquity 73(1):3-28 Miller, S. and J. Gingerich 2013 Paleoindian Chronology and the Eastern Fluted Point Tradition. In In the Eastern Fluted Point Tradition. Edited by J. Gingerich, pp 9-37. University of Utah Press, Salt Lake City.
162
Mitteroecker, P. and P. Gunz 2009 Advances in Geometric Morphometrics. Evolutionary Biology 36:235–247 Moerschfelder, F. 1981 An Archaeological Survey of the Former Townships of Rainham and South Cayuga, Regional Municipality of Haldimand-Norfolk, Province of Ontario, Ms. on file at the Archaeology Office, Ministry of Citizenship and Culture, London. 1982 Report of Archaeological License Number 82-24: A Study of Prehistoric Sites in Haldimand County. Ms on file at the Archaeology Office, Ministry of Citizenship and Culture, London. 1985 Locating the Source of Haldimand Cherts and Related Workshop Sites along Roger's Creek: An Ongoing Survey in Haldimand County. Ms. on file at the Archaeology Office, Ministry of Citizenship and Culture, London. Moran, P. A. P 1950 Notes on Continuous Stochastic Phenomena. Biometrika 37:17-23. Morgan, A.V., J.H. McAndrews and C.J. Ellis. 2000 Geological History and Paleoenvironment. In An Early Paleoindian Site near Parkhill, Ontario. Edited by C.J. Ellis and D.B. Deller, pp 9-30. Canadian Museum of Civilization, Mercury Series No. 159, Gatineau, Quebec Morrow, J.E., and A.T. Morrow 1999 Geographic Variation in Fluted Projectile Points: A Hemispheric Perspective. American Antiquity 64:215-231. Murray, A. 1997 The Ageing Maple Site: The Importance of Being Small. In Preceramic Southern Ontario. Edited by P. Woodley and P. Ramsden, pp 59-63. Occasional Papers in Northeastern Archaeology No. 9. Copetown Press, Dundas. Neiman, E D. 1995 Stylistic Variation in Evolutionary Perspective: Inferences from Decorative Diversity and Interassemblages Distance in Illinois Woodland Ceramic Assemblages. American Antiquity 60:7-36. Nelson, M. 1991 The Study of Technological Organization. Archaeological Method and Theory 3: 57-100. New Directions Archaeology Ltd. 2001 Stage 1–3 Archaeological Assessment of the Highway 401–Townline Road Interchange (W.P. 1-00-00). Report on file, Ontario Ministry of Tourism, Culture and Sport. Newby, P., J. Bradley, A. Speiss, B. Shumand and P. Leduc. 2005 A Paleoindian Response to Younger Dryas Climate Change. Quaternary Science Reviews 24:141–154 O’Brien, M.J. (ed) 2008 Cultural Transmission and Archaeology: Issues and Case Studies. Society for American Archaeology, Washington D.C.
163
O’Brien, M.J., J. Darwent, R.L. Lyman 2001 Cladistics Is Useful for Reconstructing Archaeological Phylogenies: Palaeoindian Points from the Southeastern United States. Journal of Archaeological Science 28:1115-1136. O’Brien, M.J. and R.L. Lyman 2001 Applying Evolutionary Archaeology: A Systematic Approach. Kluwer, New York. 2002 The Epistemological Nature of Archaeological Units. Anthropological Theory 2:37–57. O’Brien, M.J., B. Buchanan, M. Collard and M.T. Boulanger 2012 Cultural Cladistics and the Early Prehistory of North America. In Evolutionary Biology: Mechanisms and Trends. Edited by P. Pontarotti, pp 23-42. Springer-Verlag, Berlin. O’Brien, M.J., M. Collard, B. Buchanan and M.T. Boulanger 2013 Trees, Thickets, or Something in Between? Recent Theoretical and Empirical Work in Cultural Phylogeny. Israel Journal of Ecology & Evolution 59(2):45–61. O’Brien, M.J., M.T. Boulanger, B. Buchanan, M. Collard, R.L. Lyman, J. Darwent 2014 Innovation and Cultural Transmission in the American Paleolithic: Phylogenetic Analysis of Eastern Paleoindian Projectile-Point Classes. Journal of Anthropological Archaeology 34:100–119. Oden, N. L. and R. R. Sokal. 1986 Directional Autocorrelation: an Extension of Spatial Correlograms to Two Dimensions. Systematic Zoolology. 35: 608-617. Oswalt, W. H. 1973 Habitat and Technology: the Evolution of Hunting. Rinehart, and Winston, New York. 1976 An Anthropological Analysis of Food-Getting Technology. Wiley, New York. Overstreet, D. F. 1993 Chesrow: A Paleoindian Complex in the Southern Lake Michigan Basin. Case Studies in Great Lakes Archaeology No. 2. Great Lake Archaeological Press, Milwaukee, Wisconsin. Parker, L. R. 1986a Haldimand Chert: A Preferred Raw Material in Southwestern Ontario during the Early Holocene Period. Kewa 86:4-12. 1986b Haldimand Chert and its Utilization during the Early Holocene Period in Southwestern Ontario. M.A. Thesis, Department of Anthropology, Trent University, Peterborough. Parkins, W. 1977 Onondaga Chert: Geological and Palynological Studies as Applied to Archaeology. Master’s thesis, Department of Geology, Brock University. St. Catharines. Payne, J. H. 1982 The Western Basin Paleo-Indian and Early Archaic Sequences, unpublished B.A. thesis, Department of Sociology, Anthropology, and Social Work, University of Toledo, Toledo. Peck, J. R.G. Barreau and S.C. Heath 1997 Imperfect Genes, Fisherian Mutation and the Evolution of Sex. Genetics 145: 1171–1199.
164
Pitblado, B.I. 2003 Late Paleoindian Occupations of the Southern Rocky Mountains. University of Colorado Press, Boulder, Colorado Reyment R.A 1991 Multidimensional Paleobiology. Pergamon Press, New York. 2010 Morphometrics: An Historical Essay. In Morphometrics for Nonmorphometricians. Edited by A. Elewa, pp 9 – 26, Springer, New York Richardson, P.J., R. Boyd, R.L. Bettinger 2009 Cultural Innovations and Demographic Change. Human Biology 81(2–3):211–235. Richie, W.A. 1961 A Typology and Nomenclature for New York Projectile Points. New York State Museum and Science Service, Bulletin No. 384. Albany: The University of the State of New York. 1969 The Archaeology of Martha's Vineyard. A Framework for the Prehistory of Southern New England. Garden City, NY: The Natural History Press. 1971 The Archaic in New York. The Bulletin, Journal of the New York State Archaeological Association 52:2-12. 1979 Some Regional Ecological Factors in the Prehistory of Man. The Bulletin, Journal of the New York State Archaeological Association 75:14-22 Ritchie, W.A. and R.E. Funk 1971 Evidence for Early Archaic Occupations on Staten Island. Pennsylvania Archaeologist 41(3): 45-61 1973 Aboriginal Settlement Patterns in the Northeast. New York State Museum and Science Service Memoir 20. Albany: The University of the State of New York Roberts, A.C. 1985 Preceramic Occupations along the North Shore of Lake Ontario. National Museum of Man, Mercury Series, No. 132 Rohlf F. J., Marcus L. F., 1993 A Revolution in Morphometrics. Trends in Ecology and Evolution 8: 129-132. 2008 tpsDIG2, version 2.12. Department of Ecology and Evolution, State University of New York, Stony Brook Roosa, W.B 1963 Some Michigan Fluted Point Sites and Types. Michigan Archaeologist, 9(3): 44-48. 1965 Some Great Lakes Fluted Point Types. Michigan Archaeologist 9(3): 89-102. 1977 Great Lakes Paleoindian: The Parkhill Site, Ontario. In Amerinds and Their Paleoenvironments in Northeastern North America. Edited by W.S. Newman and B. Salwen, Annals of the New York Academy of Sciences, Vol. 288, pp. 349-354, New York. Shay, C. T. 1971 The Itasca Bison Kill Site: An Ecological Analysis. Minnesota Historical Society. Minneapolis.
165
Shennan S.J. 2000 Population, Culture History and the Dynamics of Culture Change. Current Anthropology 41:811-835. 2001 Demography and Cultural Innovation: Model and Its Implications for the Emergence of Modern Human Culture. Cambridge Archaeology Journal 11(1):5-16. 2002 Genes, Memes, and Human History: Darwinian Archaeology and Cultural Evolution. Thames and Hudson, London. Sholts, S.B, D.J. Stanford, L.M. Flores and S. Wärmländer 2012 Flake Scar Patterns of Clovis Points Analyzed with a New Digital Morphometrics Approach: Evidence for Direct Transmission of Technological Knowledge Across Early North America. Journal of Archaeological Science 39 3018-3026 Shott, M.B. 1986 Settlement Mobility and Technological Organization among Great Lakes Paleo-Indian Foragers. PhD Dissertation, Department of Anthropology, University of Michigan. University Microfilms, Ann Arbor, Michigan. 2014 Digitizing Archaeology: a Subtle Revolution in Analysis. World Archaeology 46(1): 1–9. Simons, D. B., M. J. Shott, and H. T. Wright 1984 The Gainey Site: Variability in a Great Lakes Paleo-Indian Assemblage, Archaeology of Eastern North America, 12, pp. 266-279 Simons, D. L. 1997 The Gainey and Butler Sites as Focal Points for Caribou and People. In Caribou and Reindeer Hunters of the Northern Hemisphere, edited by L. J. Jackson and P. T. Thacker, pp. 105-131. Avebury, Aldershot, England. Slice D.E 2005 Modern Morphometrics. In Modern Morphometrics in Physical Anthropology. Edited by D.E. Slice, pp. 1–45. New York: Kluwer Acad. Plenum 2007 Geometric Morphometrics. Annual Review of Anthropology 36: 261–281. Smallwood, A.M. 2012 Clovis Technology and Settlement in the American Southeast: Using Biface Analysis to Evaluate Dispersal Models. American Antiquity 77(4):689–713. Smith, P., W. Englebrecht, J.D. Holland 2010 Late-Paleoindian Archaeology at the Eaton Site, Western New York. Current Research in the Pleistocene 27:142-145 Smith, K. P., N. O’Donnell, and J. D. Holland 1998 The Early and Middle Archaic in the Niagara Frontier: Documenting the ‘Missing Years’ in Lower Great Lakes prehistory. Bulletin of the Buffalo Society of Natural Sciences 36:1–79. Soday, F.J. 1954 The Quad Site, A Paleo-lndian Village in Northern Alabama. Tennessee Archaeologist, Vol. X, No. 1, pp. 1-20.
166
Sokal, R. R. 1986 Spatial Data Analysis and Historical Processes. In Data analysis and informatics. Edited by E. Diday et al., pp 29-43. IV. North-Holland, Amsterdam. Spiess, A. E., and D. B. Wilson 1989 Paleoindian Lithic Distribution in the New England-Maritimes Region. In Eastern Paleoindian Lithic Resource Use. Edited by C. J. Ellis and J. C. Lothrop, pp 99-137. Westview Press, Boulder, Colorado. Spiess, A.E., D.B. Wilson and J. Bradley. 1998 Paleoindian Occupation in the New England-Maritimes Region: Beyond Cultural Ecology. Archaeology of Eastern North America 26:201–264. Speth, J.D., K. Newlander, A. White, A.K. Lemke, L.E. Anderson 2013 Early Paleoindian Big-game Hunting in North America: Provisioning or Politics? Quaternary International 285:111–139 Stanford, D. J., and B. A. Bradley 2012 Across Atlantic Ice: The Origin of America’s Clovis Culture. Berkeley: University of California Press. Storck, P.L. (ed) 1997 The Fisher Site: Archaeological, Geological and Paleobotanical Studies at an Early Paleo-Indian Site in Southern Ontario. Memoirs of the Museum of Anthropology, University of Michigan No. 30. Storck, P. L. 1973 Two Palaeo-Indian Projectile Points from the Bronte Creek Gap. Royal Ontario Museum, Archaeology Paper, No. 1. Toronto. 1974a The Barest Trace: New Developments in the Search for Early Man in Ontario. Rotunda 7(1):4-9. 1974b Ancient Beaches and the Trail of Early Man in Ontario. Royal Ontario Museum, Archaeological Newsletter, No. 115:4. 1979 A Report on the Banting and Hussey Sites: Two Palaeo-Indian Campsites in Simcoe County, Southern Ontario. National Museum of Man, Mercury Series No. 93. Ottawa, Ontario. 1982 Palaeo-Indian Settlement Patterns Associated with the Strandline of Glacial Lake Algonquin in Southcentral Ontario. Canadian Journal of Archaeology 6:1-31. 1984a Glacial Lake Algonquin and Paleo-Indian Settlement Patterns in Southcentral Ontario. Archaeology of Eastern North America 12:286-29. 1984b Research into the Early Paleo-Indian Occupations of Ontario: A Review. Ontario Archaeology 41:3-28. Storck, P.L. and A.E. Speiss 1994 The Significance of New Faunal Identifications Attributed to an Early Paleo-Indian (Gainey Complex) Occupation at the Udora Site, Ontario, Canada. American Antiquity, 59:121-142.
167
Storck, P.L., and J.H. Tomenchuk 1990 An Early Paleo-Indian Cache of Informal Tools at the Udora Site, Ontario. In Early PaleoindianEconomies of Eastern North America Research in Economic Anthropology. Edited by K.B Tankersley and B.L. Isaac, pp 45-93. JAI Press, Greenwich, Connecticut. Storck, P. L. and P. H. von Bitter. 1981 The Search Changes Direction: Ice Age Man in Ontario. Rotunda 14(3):28-36. 1989 The Geological Age and Occurrence of Fossil Hill Formation Chert: Implications for Early Paleoindian Settlement Patterns. In Eastern Paleoindian Lithic Resource Use, edited by C. J. Ellis and J. C. Lothrop, pp. 165–89. Westview Press, Boulder, Colorado. Tankersley K.B. 1989 A Close Look at the Big Picture: Early Paleoindian Lithic Resource Use in the Midwestern United States. In Eastern Paleoindian Lithic Resource Use. Edited by C.J. Ellis and J.C. Lothrop Westview Press, 259-293, Boulder, Colorado. Tankersley, K. B., J. D. Holland, and R. L. Kilmer 1996 Geoarchaeology of the Kilmer Site: a Paleoindian Habitation in the Appalachian Uplands. North American Archaeologist 17(2):93–113. Telford, P.G. and G.A. Tarrant 1975 Paleozoic Geology of the Dunnville Area, Southern Ontario. Ontario Division of Mines, Preliminary Map P. 988, Geological Series, Scale 1:50,000. Geology 1974. Toronto Thulman D.K. 2012 Discriminating Paleoindian Point Types from Florida using Landmark Geometric Morphometrics. Journal of Archaeological Science 39 1599 – 1607. Timmermans, S. 1999 The Southwinds site: A Late Paleo-Indian “Hi-Ho” Encampment in Middlesex County. Kewa 99:4 14. Timmins, P. 1995 Stelco I: A Late Paleo-Indian Hi-Lo Site in the Region of Haldimand-Norfolk. Kewa 95:2-22 TMHC (Timmins Martelle Heritage Consultants Inc.) 2004 Stage 4 Executive Summary, AgHb-238: Blue Box Site, AgHb-239: Snowhill Site and AgHb-240: Double Take Site, Hampton Estates Subdivision, Part of Lots 22 and 23, Concession 3, Brantford, Brant County, Ontario. Manuscript on file, Ontario Ministry of Tourism, Culture and Sport, Toronto. Tinkler K. J. and J. W. Pengelly 2004 Lake Level Changes and Aboriginal Cultural Manifestations in Areas Adjacent to and Including Niagara Peninsula. In The Late Palaeo-Indian Great Lakes: Geological and Archaeological Investigations of Late Pleistocene and Early Holocene Environments. Edited by L. J. Jackson and A. Hinshelwood, pp. 201–24. Mercury Series Archaeological Paper 165 Canadian Museum of Civilization, Ottawa, Canada.
168
Torrence, R. 1983 Time Budgeting and Hunter-Gatherer Technology. In Hunter-Gatherer Economy in Prehistory. Edited by G. Bailey, pp.11–22. Cambridge University Press, Cambridge. 1989 Re-Tooling: Towards a Behavioral Theory of Stone Tools. In Time, Energy and Stone Tools. Edited by R. Torrence, pp. 57–66. Cambridge University Press, Cambridge. 2000 Hunter-Gatherer Technology: Macro and Microscale Approaches. In Hunter-gatherers: An Interdisciplinary Perspective. Edited by C. Panter-Brick, R. H. Layton, and P. Rowley-Conwy, pp. 99–143. Cambridge University Press, Cambridge. Walthall, John A., and Brad Koldehoff 1998 Hunter-Gatherer Interaction and Alliance Formation: Dalton and the Cult of the Long Blade. Plains Anthropologist 43(165):257-273. Waters, M.R., and T.W. Stafford 2007 Redefining the Age of Clovis: Implications for the Peopling of the Americas. Science 315:1122-1126. Watson G. 2001 Report of 1999-2000 Field Work at the Hi-Lo Green Site (BdGb-2). The Ottawa Archaeologist 28(3) Weissner, P. 1983 Style and Social Information in Kalahari-San Projectile-Points. American Antiquity 48:253-276. White, A. A. 2006 A Model of Paleoindian Hafted Biface Chronology in Northeastern Indiana, Archaeology of Eastern North America, 34, pp. 29-59. 2012 The Social Networks of Early Hunter-Gatherers in Midcontinental North America. Unpublished PhD Dissertation, Dept. of Anthropology, University of Michigan, Ann Arbor, Michigan. 2013 Functional and Stylistic Variability in Paleoindian and early Archaic Projectile Points from Midcontinental North America. North American Archaeologist, 34(1):71-108. Wilhelmsen, K.H. 2001 Building the Framework for an Evolutionary Explanation of Projectile Point Variation: An Example from the Central Mississippi River Valley. In Posing Questions for a Scientific Archaeology. Edited by T. L. Hunt, C. P. Lipo, and S. L. Sterling, 97–144. Bergin & Garvey, Westport. Winterhalder, B. P., F. Lu and B. Tucker 1999 Risk-Sensitive Adaptive Tactics: Models and Evidence from Subsistence Studies in Biology and Anthropology. Journal of Archaeological Research 7(4): 301-348. Wobst, H. M. 1978 The Archaeo-Ethnology of Hunter-Gatherers or the Tyranny of the Ethnographic Record in Archaeology. American Antiquity, 43(2):303-309
169
Woodley, P. 1997 The Witz (AhHa-80) and Koeppe II (AhHa-157) Sites, Ancaster and the Hi-Lo Occupation of Southern Ontario. In Preceramic Southern Ontario. Edited by P. Woodley and P. Ramsden, pp 147-169. Occasional Papers in Northeastern Archaeology No. 9. Copetown Press, Dundas. 2004 Fowler Site: a Holcombe Camp near Lake Simcoe, Ontario. In The Late Palaeo-Indian Great Lakes: Geological and Archaeological Investigations of Late Pleistocene and Early Holocene Environments. Edited by L. J. Jackson and A. Hinshelwood, pp. 163–200. Mercury Series, Paper No. 165, Canadian Museum of Civilization, Ottawa, Canada. Wormington, H.M. 1957 Ancient Man in North America. In: Popular Series, No. 4. Denver Museum of Natural History. Wray, C. F. 1948 Varieties and Sources of Flint Found in New York State. Pennsylvania Archaeologist 18(1):25–45. Wright, H.T. and W.B. Roosa 1966 The Barnes Site: A Fluted Point Assemblage from the Great Lakes Region. American Antiquity 31(6): 850-860. Wright, P and G. Watson 1999 Report of 1998 Field Work at the Hi-Lo Green Site (BdGb-2) and the Multi-Component Archaic Cockrill Site (BdGb-3) in Eastern Ontario. The Ottawa Archaeologist 27(3) Wright, J.V. 1978 The Implications of Probable Early and Middle Archaic Projectile Points from Southern Ontario. Canadian Journal of Archaeology 2:59-78. 1995 A History of the Native People of Canada, Volume 1 (10,000-1,000 BC). Mercury Series No. 152, Canadian Museum of Civilization, Ottawa. Yu, Z. 2000 Ecosystem Response to Lateglacial and Early Holocene Climate Oscillations in the Great Lakes Region of North America. Quaternary Science Reviews 19: 1723-1747