Mesolithic Northern England: Environment, population and settlement 9781841710068, 9781407319155

Table of contents :
Front Cover
Title Page
Copyright
Dedication
Preface
Table of Contents
LIST OF FIGURES
LIST OF TABLES
Chapter One: Population and Adaptation
Chapter Two: Interpreting Site Distributions
Chapter Three: Working 'Up' from Resources
Chapter Four: Ecological and Ethnographic Analogies
Chapter Five: Models of Changing Environments
Chapter Six: The Implications
Conclusions
Appendices
Bibliography

Citation preview

BAR 283 1999  SPIKINS  MESOLITHIC NORTHERN ENGLAND

Mesolithic Northern England Environment, population and settlement

Penny Spikins

BAR British Series 283 9 781841 710068

B A R

1999

Mesolithic N orthem England Environment, population and settlement

Penny Spikins

BAR British Series 283 1999

Published in 2016 by BAR Publishing, Oxford BAR British Series 283 Mesolithic Northern England © P Spikins and the Publisher 1999 The author's moral rights under the 1988 UK Copyright, Designs and Patents Act are hereby expressly asserted. All rights reserved. No part of this work may be copied, reproduced, stored, sold, distributed, scanned, saved in any form of digital format or transmitted in any form digitally, without the written permission of the Publisher.

ISBN 9781841710068 paperback ISBN 9781407319155 e-format DOI https://doi.org/10.30861/9781841710068 A catalogue record for this book is available from the British Library BAR Publishing is the trading name of British Archaeological Reports (Oxford) Ltd. British Archaeological Reports was first incorporated in 1974 to publish the BAR Series, International and British. In 1992 Hadrian Books Ltd became part of the BAR group. This volume was originally published by Archaeopress in conjunction with British Archaeological Reports (Oxford) Ltd / Hadrian Books Ltd, the Series principal publisher, in 1999. This present volume is published by BAR Publishing, 2016.

BAR

PUBLISHING BAR titles are available from:

E MAIL P HONE F AX

BAR Publishing 122 Banbury Rd, Oxford, OX2 7BP, UK [email protected] +44 (0)1865 310431 +44 (0)1865 316916 www.barpublishing.com

Mesolithic Northern England Environment, Population and Settlement

Penny Spikins

Departamento de Arqueologfa Facultad de Ciencias Naturales Museo de la Plata 1900 La Plata Argentina Department of Archaeology The Quadrangle University of Newcastle-upon-Tyne Newcastle NE1 ?RU (UK)

email: [email protected]

ii

To Horacio, for limitless patience and enthusiasm.

iii

iv

Preface

Focusing on evidence from northern England, this volume addresses the idea of gradual population increase and related concepts which structure our understanding of Mesolithic settlement strategies. One of the key aims is to illustrate the influence which our preconceptions about Mesolithic societies (preconcieved ideas about population changes or settlement patterns for example) can have in structuring our understanding of the archaeological evidence. By critically assessing both the nature of the archaeological and environmental evidence for Mesolithic adaptations, and the limitations of traditional approaches, it is suggested that many of our most basic interpretations of the period are in fact often poorly grounded assumptions. The concept of gradual population increase throughout the Mesolithic provides the 'thread' of the discussion, with other issues such as resource exploitation and settlement patterns brought in as related concepts. We then offer one possible different approach, acknowledging the importance of ecological changes in a large scale model of changing vegetation, but attempting to avoid static and deterministic interpretations. Potential adaptations to the ecological changes suggested by the model provide alternatives to traditional explanations for many of the observed patterns in the archaeological record. Through a final re-assessment of the notion of gradual population increase and related concepts, the way is paved for more dynamic and realistic interpretations of Mesolithic human adaptations.

Chapter one introduces the concept of gradual population increase and the reasons for selecting northern England as an appropriate case study. The effects of present biases and the history of investigations on interpretations of the archaeological record (chapter two) and the evidence for available resources and their exploitation (chapter three) are discussed in turn. The use of often simplistic ecological and ethnographic analogies to interpret this evidence is assessed in chapter four. One possible 'way forward', a 'coarse grain' interpretation of spatial changes in vegetation zones, is described in the next two chapters, with Chapter five describing the construction of a Geographical Information System based model of changing vegetation in northern England, and chapter six, the possible human adaptations to these changes and the implications for the concept of gradual population increase. Wider implications of the findings are considered in the Conclusions. Although the results of recent fieldwork in the region are discussed where appropriate, the volume is not intended as a summary of the evidence for Mesolithic occupation but rather as a 're-thinking' of popular preconceptions and accepted means of interpreting Mesolithic activities. The two people I am most indebted to in producing this volume are my husband, Horacio Ayestaran, and my PhD supervisor, Todd Whitelaw. Both have showed endless patience to listen and respond to my ideas. Academic advice from those in the Archaeology Departments at Cambridge, particularly Paul Mellars and Colin Shell, at Newcastle, particularly Geoff Bailey, and also Pete Rowley-Conwy at Durham and Rob Young at Leicester have also contributed to the formulation of many of my ideas. Specialists in Earth Sciences and Botany have helped out with what seemed at first intractable disciplines and terminology, and I would particularly like to thank Keith Bennett, now Professor at the University of Uppsala, Sweden. I would also like to thank the staff at West Yorkshire Archaeology Service, where I was first introduced to the Mesolithic of northern England, particularly Dave Berg, Bob Yarwood and Jenny Marriott, and this organisation alongside English Heritage and the National Trust for supporting my fieldwork project. I must also thank NERC for the financial support that made the production of this thesis possible, plus a Sir James Knott Fellowship at Newcastle University and, during the ultimate amendments, a Leverhulme Trust Award at La Plata University, Argentina. I am also grateful to those individuals who have been kind enough to allow me access to their research and site records, such as Chantal Conneller, Tina Dudley, John Castleford, Bob Yarwood, Jenny Marriott, Andy Myers, Norman Redhead and Sue Whitely. I would also like to thank the members of the Garrod Lab and the Palaeolithic/Mesolithic V

group at Cambridge for providing much needed encouragement and a stimulating environment in which to work on my thesis - the basis of this volume, and more recently to staff at the Archaeology Department in Newcastle, particularly Nick Winder and Ruth Charles. Lastly, I would like to thank Celina Migliaro for advice on page layout and graphic design . For the illustrations, I would like to thank P. Rowley-Conwy (for permission to use figures 4.4 and 4.9), C. Bonsall (for permission to use figure 4.5), H. Birks (for permission to use figure 5.3), R. Jacobi (for permission to use figures 2.4 and 2.5), C. Tolan-Smith (for permission to use figure 2.2), I. Simmons (permission to use figure 4.1, figure 4.2, and table 4.1,), M. Jochim (for permission to use figure 4.3 and table 4.2), T. Atkinson, K. Briffa and G. Coope, and McMillian publications (for permission to use figure 5.1), and the Royal Geographical Society (for permission to use figure 4.6, photo by W.S. Barclay c.1901-3). It is hoped that the reader finds the text approachable and that the discussions provide a means of re-thinking some of our approaches to Mesolithic societies and stimulating the development of new, and perhaps more dynamic interpretations.

Contents

CHAPTER ONE:

Population and Adaptation

Abstract Common Adaptations Population Increase Defining a Strategy

CHAPTER Two:

Interpreting Site Distributions

Abstract Variations over Time Variations in Space Large Scale Distributions Regional Scale Distributions Local Scale Distributions The Effects of Bias The Temporal Patterning of Sites Biases in Spatial Distributions Types of Bias Marsden Moor - A Detailed Study. Conclusions

1 1 2 3 5

7 7 8 9 9 11

13 15 15 17 17 22

28

CHAPTER THREE:

Working 'Up' from Resources

29

Abstract Introduction Subsistence Resources The Spread of Flora and Fauna into the British Isles Large Land Mammals Discussion Small Mammals Plant Foods Other Uses of Plant Resources Discussion Marine Resources Large Sea Mammals Marine Fish Migratory Birds The Role of Different Coastal Resources Lake Fish and Inland Birds What was the Role of Different Resources?

CHAPTER FOUR:

29 30 32 32 32 34 35 37 40 41 42

43 44

46 47 48

49

Ecological and Ethnographic Analogies

Abstract Artefacts, Eco-facts and Ethno-facts The Ecological Approach The Limitations to Current Ecological Models of Subsistence and Settlement Population The Ethnographic Approach Population Settlement Systems Models of Inland Groups Coastal Complexity Models Limitations to Current Ethnographic Models of Subsistence and Settlement Conclusions

viii

51 51 52 53 59

60 62 63

64 65

67 68 75

CHAPTER FIVE:Models of Changing Environments Abstract Introduction Mesolithic Environments The Nature of Terrestrial Environments Approaches to Large Scale Changes in Vegetation Ecological Determinants of Woodland Distributions Predicting Past Vegetation - In Principle Evidence for the Determining Factors The Resolution of the Model Constructing The Model Predicted Patterns of Changes in Vegetation The Relationship of the Phases to Archaeological Chronology The Pattern of Changing Woodland Types The Limitations of the Model Potential Improvements and Applicability Conclusions

77 77

78 78 78 79 79

80 80

84 85 89 96 96 100 104 104

105

CHAPTERSIX: The Implications Abstract Introduction Changing Resources Possible Adaptations The Implications of Changing Environments for Changes in Available Resources The Ecological Basis The Abundance of Resources The Diversity of Resources Combining Different Influences on Resource Abundance The Environmental Changes The Effect of Environmental Changes on the Distribution of Resources Adaptations to Ecological Changes A Shift in the Location of the Most Abundant Resource Zones The Northward Movement of Environmental Zones The Fragmentation of Resource-Rich Environments Wider Implications The Colonisation of Northern Regions The Mesolithic-Neolithic Transition Population Increase Re-assessed Archaeological Evidence for Increases in Population Ecological Changes Other Considerations - Adaptability Conclusions

105 106 106 106 107 107 107 108 108 109 109 112 112 117 119

121 121 122 123 123 124 124 125

CONCLUSIONS

127

APPENDICES

130

BIBLIOGRAPHY

136

ix

Figures CHAPTER ONE: Figure 1.1 Figure 1.2

Northern England, topography and regions. The structure of this volume.

CHAPTER Two: Figure Figure Figure Figure Figure

2.1 2.2 2.3 2.4 2.5

Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure

2.6 2.7 2.8 2.9 2.10 2.11 2.12 2.13 2.14 2.15 2.16 2.17 2.18

Figure Figure Figure Figure Figure Figure

2.19 2.20 2.21 2.22 2.23 2.24

Population and Adaptation 5 6

Interpreting Site Distributions

Scales of Spatial Patterning. Radio-carbon dated sites in the British Isles (after Smith 1992). Mesolithic sites in northern England. Potential 'social territories' according to assemblage types in England (after Jacobi 1979: 68). Concentrations of Early Mesolithic sites in the Pennines and Yorkshire Wolds (after Jacobi 1978: 304). Sites in the southern Pennines and areas of most severe peat erosion. Mesolithic sites on Marsden moor (data from Stonehouse 1990). Early and Late Mesolithic sites on Marsden Moor. Elevation of sites and non-sites in the Pennines (after Spikins 1995c: 95). Numbers of artefacts on Mesolithic sites in the Pennines. Mainland Britain radio-carbon dates v. Northings, with Northern England typology. Calibrated dates a) for the British Mesolithic and b) for northern England. Main centres of population in northern England. Numbers of recorded sites by decade in the Central Pennines. Deposits affecting the visibility of sites in south-east northern England. West Yorkshire Mesolithic Project - excavations on Marsden Moor. Vegetation Survey, main vegetation types. Vegetation Survey, types of erosion. North-facing slope of Dean Clough. Central Plateau. South-facing slope of Dean Clough. Model of Erosion types. Recorded marginal peat face erosion in the southern Pennines and the distribution of recorded sites. Model of marginal face erosion in the Pennines and the distribution of recorded sites.

8

9

10 10 11

12 13 13 14 15 16 16 18 19 19 22 23 23 24 24 24 25 26 27

CHAPTERTHREE:Working 'Up' from Resources Figure 3.1

Sites in northern England mentioned in chapter three.

31

CHAPTERFOUR: Ecological and Ethnographic Analogies Figure 4.1 Figure 4.2 Figure 4.3 Figure Figure Figure Figure Figure Figure Figure

4.4 4.5 4.6 4.7 4.8 4.9 4.10

Proposed Late Mesolithic resource availability of north-east Yorkshire (Simmons 1979). Simmons' (1996: 215) 'most likely' model of Mesolithic settlement. Model of seasonal resource exploitation in the Mesolithic of south-west Germany (Jochim 1976: 115). Proposed resource availability schedule for the Danish Ertebjljlle (Rowley-Conwy 1984). Proposed Late Mesolithic resource availability in the Eskmeals area (Bonsall 1981: 466). The Selk'nam of Tierra del Fuego: Hunter-gatherers in a forested environment. The accepted structure to Mesolithic settlement. Archaeological evidence for Mesolithic settlement patterns. Resource availability for maritime hunter-gatherers (Rowley-Conwy 1986). A 'better explanation' for patterns in recorded sites?

53 54 56 58 58 62 65 66 67 71

CHAPTERFIVE: Models of Changing Environments Figure 5.1

Temperature changes in the British Isles reconstructed from beetle remains, after Atkinson, Briffa and Coope (1987: figure 2).

Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure

8180/

5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12 5.13 5.14 5.15 5.16 5.17 5.18 5.19

8160

ratio recorded in the GISP2 ice core (Meese et al 1994; Stuiver et al 1995). The spread of oak in the British Isles (after Birks 1989: 512). The construction of the model. Assigning probable dominant vegetation. Soil preferences base map. Topography base map. Vegetation succession used in the model. Model of Probable Dominant Woodland Types at 10,000bp (above) and 9,750bp (below). Model of Probable Dominant Woodland Types at 9,500bp (above) and 9,000bp (below). Model of Probable Dominant Woodland Types at 8,500bp (above) and 8,000bp (below). Model of Probable Dominant Woodland Types at 7,500bp (above) and 7,000bp (below). Model of Probable Dominant Woodland Types at 6,500bp (above) and 6,000bp (below). Model of Probable Dominant Woodland Types at 5,500bp (above) and 5,000bp (below). Model of Probable Dominant Woodland Types for the 'Early Mesolithic'. Model of Probable Dominant Woodland Types for the 'Initial Late Mesolithic'. Model of Probable Dominant Woodland Types for the 'Terminal Late Mesolithic'. Terminal Late Mesolithic (6,000bp) model based on the present distribution of alluvial gley soils. Early Mesolithic (9,500bp) model with rapid early Holocene temperature rise.

XI

81 82 83 85 86 87 87 88 90 91 92 93 94 95 97 98 99 101 102

CHAPTER SIX:

The Implications

Figure 6.1

Sequence of changes in vegetation patterns in Mesolithic northern England.

109

Figure 6.2 Figure 6.3 Figure 6.4

Changes in resource availability with vegetational changes in the Mesolithic. Changes in upland and lowland resource zones in the Mesolithic. Early and Late Mesolithic sites in the Central Pennines,

110 113

with vegetation model for the Initial Late Mesolithic. Distribution of Early Mesolithic radio-carbon dated sites in northern England.

115

Figure 6.5 Figure 6.6

Distribution of Late Mesolithic radio-carbon dated sites in northern England.

116 116

Figure 6.7

Alternative explanations for increases in recorded Mesolithic sites.

123

Tables Table 2.1

Numbers (thousands) of sheep in the North and West Riding of Yorkshire (after Evans 1992).

18

Table 4.1

Seasonal resource use model for England and Wales (Simmons 1996: 199).

54

Table 4.2 Table 4.3

Percentage contribution to diet of major resource groups (Jochim 1976: 108). Resource exploitation seasons (Jochim 1976: 116).

55 56

Table 4.4

Contributions of groups of resources to diet for Canadian boreal forest groups (Price 1973 after Rogers 1966).

64

xii

CHAPTER ONE

Population and Adaptation

ABSTRACT In this chapter the concept of gradual increase in population throughout the Mesolithic of Western Europe is introduced. The context of gradual population increase as one of the common adaptations in the Mesolithic is described, alongside the wider context of general ideas about gradual increases in populations amongst all hunter-gatherer populations. Gradual population increase is taken as an example of a common preconception brought to bear on Mesolithic societies, the validity of which will be brought into question by assessing the different lines of evidence for interpretations in the following chapters. The history and evidence for conclusions about population increase in northern England, and the suitability of this region as a case study are outlined in this chapter, leading on to the archaeological evidence for population and settlement discussed in chapter two.

In this volume, we will be focusing on the concept of gradual population increase throughout the Mesolithic, as an example of one of the most commonly accepted generalisations associated with Mesolithic populations. A discussion, essentially a critique, of the concept of population increase forms the 'core' of the volume. However, it is hoped that by addressing in detail the different lines of evidence and types of interpretations which contribute to this concept, we will also reveal the weaknesses in many of our preconceptions about Mesolithic societies. Building on the problems defined, it is hoped to view Mesolithic societies with a new perspective, and to suggest possible new approaches to interpretations.

COMMON ADAPTATIONS The idea of adaptation, that is, 'the fact, act, process or result of adapting (to adapt - to make fit or suitable): ... adjustment' 1became important within archaeology through perspectives brought in by the New Archaeology and in particular through cultural ecology (Trigger 1989: 294-303; Kelly 1995: 41-50). Discussions of adaptations have often been associated with a narrow environmentally deterministic standpoint, through which environmental characteristics determine aspects of society - such as population densities and settlement structures. Thus, at an extreme, the cultural ecological approach to adaptation viewed almost all cultural changes as adaptations to environmental change, and as such, culture as largely predictable by environment. However, today most Mesolithic researchers tend to interpret adaptation in its broader sense, as an adjustment to both environmental and social changes. In this sense, for example, the rise of social complexity in the Mesolithic is often seen as an adaptation, but to factors such as increasing population stress or a demand for certain goods rather than directly to a specific environmental change.

Gradual population increase is often seen as one of the common adaptations that become evident in the Mesolithic, apparently related to changing environmental conditions. Thus in this chapter the role of the concept of adaptations, in particular the common adaptations which appear to unite almost all Mesolithic societies, in structuring our interpretations of the Mesolithic is briefly discussed. Gradual population increase is introduced in the context of these accepted adaptations, and within its broader context as a common preconception brought to bear on hunter-gatherer societies. The background to interpretations of gradual population increase in the study region - northern England is also outlined.

Any adaptation, such as, for example, an increased consumption of shellfish, when other coastal resources become scarce - perhaps due to environmental changes or overexploitation, may be very locally specific. However, common adaptations play a particularly important role in the study of the Mesolithic. Broadly speaking, common adaptations are essentially widespread changes in aspects of society - population or settlement for example - which appear to be related to changes in common social or environmental contexts. Of these, changes in the exploitation of resources (specifically an apparent diversification and specialisation of the resource base), gradual population increase, and rising social complexity are the three main themes. One reason why these common adaptations are an important area for Mesolithic studies is straightforward - the Early Holocene was a time of significant environmental changes, such that an obvious focus of study becomes the common response of human societies to widespread changes. However, there is also a much more complicated reason why common adaptations play such a key role. The roots of this explanation lie in common perceptions of Mesolithic research as a discipline. Essentially, a paucity of distinctive Mesolithic material culture or achievements (with diminutive microliths often seen as an appropriate symbol of the period, Clark 1978:3) has been widely acknowledged as leading to a poor public perception of the Mesolithic for some time (Zvelebil 1986; Rowley-Conwy 1986). Price thus described the wider academic perspective of the Mesolithic in late 80s as having a 'certain disregard for the period as one of relatively little importance' (1987: 229). Paradoxically, this poor popular perception of the period has generated a certain atmosphere of camaraderie amongst researchers within the discipline, which whilst promoting much collaborative work 1

2

Chambers English Dictionary, Cambridge University Press 1988, p14

has also led to a certain lack of criticism of accepted ideas. On the other hand however, it has also resulted in a search for what might be distinctive about the Mesolithic, and a resultant emphasis on adaptations to environmental changes as a major distinguishing feature of the period. Newell even suggests that the Mesolithic itself be defined as an 'adaptation ... to the rapid ecological changes which marked the Pleistocene-Holocene boundary' (1984: 71). Thus, in many respects, the very identity of the Mesolithic as a period could be said to be built upon the common adaptations of Mesolithic societies. As a common adaptation, population increase sits between changes in the exploitation of resources, and an increase in social complexity (for many, not only thematically, but also causally). Both changes in resource exploitation and the rise of social complexity have been discussed in detail elsewhere (Price and Brown 1985; Zvelebil 1986; Rowley-Conwy 1983; 1984; 1986; Rowley-Conwy and Zvelebil 1989; Mithen 1990). The former is typified by the notion of a diversification of the resource base (to include a range of avian and aquatic resources (Binford 1968: 317), including inland fish (LeGall 1996), as well as an increased range of plant foods (Zvelebil 1994). However, with rising sea-levels submerging what might exist of previous evidence for marine exploitation, the precise timing of such a diversification is difficult to define. Likewise, evidence for more specialised exploitation may also be affected by a lack of previous evidence. The evidence for rising social complexity, in the form of large-scale, apparently permanent settlement and cemeteries (see for example O'Shea and Zvelebil 1984; Larsson 1989) is however clear-cut, although extremely localised (particularly in certain coastal areas). Within the apparent changes taking place, population increase clearly plays a important role, particularly since there appears to be unambiguous evidence for gradually increasing populations in both inland and coastal groups.

POPULATION INCREASE

A gradual and progressive increase in population has many implications (as described by Constandse-Westermann and Newell, quoted on the opening page of this chapter). Firstly, population rise has traditionally been linked to a rise in social complexity (Testart 1982; Perlman 1980; RowleyConwy 1980, 1983, 1986; Rowley-Conwy and Zvelebil 1989; Price and Brown 1985), although, as Keeley (1988) points out, population has become somewhat less fashionable as a causal factor in recent years. Nonetheless, Keeley's analysis does show a clear statistical relationship between population pressure (in relation to resources) amongst known hunter-gatherers and social complexity. The relationship between population and complexity is not the theme addressed here, but arguments for such a relationship clearly illustrate the importance of considering population changes in discussions of rising complexity. Secondly, even in those regions where any major change in social complexity is not evident (such as in northern England), population changes are commonly associated with the adoption of agriculture (Flannery 1969; Hamer 1970; Smith and Young 1972; Cohen 1977; Hassan 1978; MacNeish 1977; Binford 1968). Binford (1968) for example sees pressures from population increase and movements of populations into areas already 'at the resource limit' as driving the intensification of resource production and the adoption of agriculture. In his model population pressure either results from increases of sedentary communities at coastal locations or from increases in inland groups placing pressure on coastal resources. Again, particularly since the 1970s, many other factors have also been put forward, but population rise still remains a 'major player', and population an important part of the context of increased intensification of resources, if not the cause. In almost every case, population increase is also linked to other processes which relate to almost all aspects of society, such as changes in settlement systems, a decrease in territory sizes and increased territoriality. For Mesolithic Western Europe, the evidence for gradual population increase during the Mesolithic appears convincing. The ecological explanation for population expansion is found within the apparently developing wealth of abundant and storable marine resources at the coast, and a gradual transition from a pine type forest to oak inland, which would represent, in crude terms, an 'improvement' of woodland resources for hunter-gatherers, with more undergrowth vegetation to support large game mammals. Myers (1989: 89), states that 'the period between 7000 and 6000bc also coincides with the sequence of major vegetational changes during which the pine and birch forest was being replaced with an increasingly diverse mixed deciduous woodland cover and dense understorey vegetation.'

Archaeological evidence supports the ecological perspective. Firstly, there are changes in the numbers of dated sites with each progressive period in many regions. In the Northwest European Plain, Newell (1973: 408) documents increases in the numbers of sites, with a threefold increase in the Late Mesolithic. The same pattern is also evident elsewhere. In the British Isles as a whole, Smith (1992b) demonstrates an increase in the numbers of radio-carbon dated sites. In England and Wales, Jacobi (1976), Morrison (1980: 136) and Myers (1986) also document clear increases in the numbers of sites which can be typologically assigned to the Early or Late Mesolithic.

population controls. Caldwell et al. (1987) for example questioned the ubiquity and effectiveness of practices such as infanticide in limiting population increase. Kelly (1995: 239 , after Blurton Jones and Sibly 1978; Blurton Jones 1986, 1987) also noted that the wide birth spacing noted amongst the Jo'houansi (!Kung) actually functioned to maximise overall reproductive success in conditions of high mobility (and hot arid climates). Put simply, hunter-gatherers may have fewer offspring than the 'biological potential' but quite 'low' birth rates (perhaps five offspring per generation) can produce marked increases in conditions of low infant mortality. The resultant changes in population may be slow on anthropological timescales, but are extremely rapid on archaeological ones. Rather than being characterised by low birth rates and low mortality as the 'affluence' model might imply, high birth and mortality rates appear to be typical (Caldwell et al 1987). This change in perspective is important as it implies that as a species, humans should be sensitive to changes in resources abundance with population numbers capable of marked changes, certainly in timescales of millennia. Rapid population adjustments and oscillations is not however what the archaeological record (nor the perception of Mesolithic adaptations) appears to imply.

A further source of evidence comes from a decrease in the size of 'style zones' through time. Since particular styles of items (tools, shelters, clothing) are often associated with a 'tribe' or 'maximal band' (Wiessner 1983; Sackett 1982; Wobst 1974; 1976), the appearance and increase in the numbers of identifiable regional industries has been interpreted in terms of increasing numbers of these bands (which would each cover a smaller area) and increases in absolute population numbers. Thus Vang Petersen (1984) documents the development of clearly delimited zones of flint flake axe heads in the Late Mesolithic of Eastern Denmark, these zones being only about 45 km in diameter. Price (1980: 220) also suggests a trend towards smaller social territories in the North European Plain, documenting a change from three distinct 'technocomplexes' in the Late Upper Palaeolithic to 8-10 distinct groups in the Early Mesolithic and over 15 identifiable groups in the Late Mesolithic. Likewise, Rozoy (1988) notes the appearance of 20 distinctive groups in the Late Mesolithic of France, and in Britain Jacobi (1979) suggests the appearance of regional social territories after 7 ,000bp in England on the basis of more regionalised artefact distributions (with smaller style zones). This increasing regionalisation in the British material has also been confirmed by other studies (such as Care 1982; Myers 1986).

The idea of gradual population increase throughout the period appears to be an appropriate notion through which to call into question our concepts of the Mesolithic. To tease out the weaknesses (and strengths) of our present perceptions and means of interpretation, we will focus on a specific region - that of northern England. Then we can address in detail the varying lines of evidence for population change and for related concepts such as changes in settlement patterns or the evidence for territoriality.

Population Increase in northern England Northern England has been chosen as a suitable region in this study for several reasons. For one thing, there has been a long history of research into the Mesolithic occupation of northern England. Northern England has been the focus of, or included within, several key publications including Mellars (1976), Jacobi (1976; 1979), Myers (1986; 1987; 1989) and Simmons (1996). There are clear increases in the numbers of Mesolithic sites whether recorded typologically (Jacobi 1976; Morrison 1980; Myers 1986) or by radiocarbon dating (Smith 1992), and clear evidence for increasing regionalisation (or a reduction in the size of territories) (Jacobi 1977; 1979; Myers 1986). Moreover, several publications and interpretations have drawn on the idea of population increase, specifically Jacobi (1976), Smith (1992b) and gradual population increase is also a component of adaptations described by Myers (1986; 1989).

Justifications for slow but inexorable population rise amongst hunter-gatherers generally also come from ethnographic literature. Discussions at the Man the Hunter conference, Chicago (Lee and DeVore 1968) in particular inspired a global model of hunter-gatherers as consciously controlling population numbers. Thus infanticide and long birth intervals (maintained through various means such as the suppression of ovulation through extended breast feeding) are widely accepted as mechanisms which in the past served to keep population stable or much limit potential increases. These mechanisms have been interpreted as a vital factor which allow 'primitive affluence' (Sahlins 1972), by keeping populations well below the carrying capacity of the environment and preventing the 'boom and bust' cycles which appear to characterise population numbers of other species.

Northern England is also a region which today illustrates a number of different environmental zones (see figure 1.1) making it appropriate for considering adaptations to different environments. Much of this environmental variability is influenced by topographic variability, from the high uplands of the Lake District to the mid-uplands of the Central Pennines and North York Moors and the lowlands. Additionally there are different geological conditions and soils (such as limestone in the Southern Pennines, granites in the Lake District), different climates (with more rainfall in the west and the uplands), and differing present land use

These latter generalisations from ethnographic studies have however recently been called into scrutiny. A closer consideration of ethnographic literature revealed little secure evidence for the conscious 'dampening' of long-term population increase amongst hunter-gatherer societies. Thus, a number of authors have questioned the supposed anthropological evidence for widespread and effective 4

In considering gradual population increase as an adaptation any population changes must be taken from after the first influx of Early Mesolithic colonists - i.e. from the earliest settled Early Mesolithic occupation to the latest Late Mesolithic. DEFINING A STRATEGY

Figure 1.1

In what follows, the suitability of the concept of gradual population increase throughout the Mesolithic of northern England will be addressed in detail. The strategy used will be to begin by considering the evidence for Mesolithic adaptations (figure 1.2). First, the 'top-down' evidence for settlement systems and population, working from the evidence left behind by these activities, then, the 'bottom up' evidence for resources and resource exploitation from the nature of resources themselves and how they might have been exploited, and finally, the means of interpreting these two elements, using analogies with modem environments and modem hunter-gatherer populations. Following these considerations, a 'fresh approach' to the question of gradual population increase and of other potential adaptations, based on models of spatial changes in resource environments, will be developed. The models used will aim to provide a link between general adaptations and the local or regional archaeological record by being based on regional environments. On the basis of this new approach, the question of gradual population increase will be re-addressed.

Northern England, topography and regions.

practices influencing site recovery, such as upland peat and lowland arable farming. The scale of northern England (approximately 180km by 240km) is also an appropriate one for considering large scale hunter-gatherer settlement systems.

Structure In this chapter, the importance of the concept of gradual population increase, both as a means of structuring our understanding of the evidence for other changes in societies, and as a means of defining the Mesolithic as a period, has been discussed. The archaeological evidence for past population and adaptations, and the biases operating on this record, are considered in Chapter two (see figure 1.2). Unfortunately, the influence of biases on interpretations of site patterning, taken both temporally and at three different spatial scales, are found to be pervasive and the potential for using this record alone as a means of understanding adaptations is found to be poor. Consequently, the evidence for available resources and their use, as a basis for 'building up' and understanding of subsistence practices and settlement patterns is addressed in Chapter three. No obvious resource provides the 'key' to subsistence practices, and Chapter four thus considers the means of interpreting the evidence from both the available resources and the archaeological record, using analogies with modem environments and populations. However, the use of these analogies, the main means of understanding Mesolithic societies, is found to be much biased by simple preconceptions

The timespan of Mesolithic occupation in this region is defined here by the appearance of the earliest sites characterised by microliths (in this case Star Carr, at 9,700 ± 160 bp, Day and Mellars 1994) and the latest (in this case March Hill Trench B, at 5190 ± 45 bp (OxA-6306))2. The former date appears about 500 years after that of 'Upper Palaeolithic' sites, which are in any case only recorded in the south of the region. In fact, the period between the Upper Palaeolithic and Mesolithic in northern England coincides with the Younger Dryas cold phase, and may possibly have been one of depopulation. In contrast, the latter date is clearly later than many Neolithic sites in the region, (although the relationship between Mesolithic and Neolithic populations is largely beyond the scope of the questions addressed here). In effect, if the region was unoccupied prior to the first Mesolithic occupation then populations must have increased by the Neolithic. The idea of gradual population increase through time is however largely separate from the issue of any initial influx of populations at the start of the Mesolithic. 2

Having discussed the evidence and the range of interpretations of the Mesolithic occupation of northern England, it becomes clear that many of our common perceptions of the Mesolithic in this region are much influenced by biases and preconceptions. Moreover, almost all approaches tend to ignore potential variability in both environments and settlement patterns. It is concluded that a new approach, aiming to address questions of change and

To convert dates to 'real years' by calibrating can be confusing as many of the dates for the Mesolithic occur on radio-carbon plateaux and can have several possible 'means' (the calibrated date, using the marine coral calibration curve and CALIB 3.0, for 9,500bp for example has three means - 10540BP, 10510BP and 10480BP - each date associated with a different probability that it may be closest to the 'real' date). For clarity radiocarbon years are used, with calibrated dates added where the actual length of a period is considered. 5

common adaptations and the role of gradual population increase

the archaeological record for population and settlement

CHAPTER TWO

CHAPTER THREE

Interpreting Site Distributions

Working 'Up'from Resources

the evidence for resource availability and use

CHAPTER FOUR

Ecological and Ethnographic Analogies

a new approach to ecological changes

the implications for gradual population increase and other Mesolithic inland adaptations

Figure 1.2

The structure of this volume

sensitive to regional environmental variability (and potential variability in settlement systems) is needed. Chapter five thus describes one such potential 'new approach', a way of moving forward from the present level of interpretations, through the development of an alternative model of changes in terrestrial environments (applied to northern England). Chapter six concludes with the implications of these changes for the idea of gradual population increase, and suggests alternative potential adaptations which may characterise inland groups. The wider relevance of this research for our understandings of some of the key concepts that structure our ideas about the Mesolithic is addressed in the Conclusions. If there is a unique approach that structures this research, it is the 'spatial' or 'geographical' perspective of the work, applied to both present biases, and past ecological changes. The use of Geographical Information Systems (GIS) has played an important role in this respect, but since it is the approach, rather than the technique, which is important (and almost all analyses included could have been carried out without the use of GIS) the workings, methods and limitations of these systems are not discussed (they are in any case amply covered in several recent publications).

interpretations using analogies with modern environments and hunter-gatherer groups

CHAPTER TWO

Interpreting Site Distributions

ABSTRACT The first source of evidence for the Mesolithic occupation of northern England has traditionally been the archaeological record. In this chapter, the archaeological evidence for past population and settlement in northern England, and how it has been interpreted, the 'top down' approach, are considered. In theory, evidence for changes through time in the numbers of Mesolithic sites can potentially tell us about population changes. Patterns through space, the spatial distribution of sites, on the other hand, have been interpreted in terms of processes ranging from large scale distinctions between potential social territories, to settlement patterns, to the way in which local landscapes were exploited, all of which are important as the contexts for changes in population. It is demonstrated here that none of these distributions are free from bias, and different types of biasing factors have different effects at different scales. Unfortunately, it even appears that, on closer scrutiny, the evidence for the Mesolithic occupation of northern England is much more biased than it might at first appear. Because of this, commonly accepted interpretations may be based on 'false patterns' occurring at different scales. The implications of these 'false patterns' are discussed. Although there may be some genuine patterning which relates to Mesolithic activities, this patterning is difficult to determine and not necessarily explainable through currently accepted models. Clearly the archaeological evidence, as it stands, is insufficient for a 'top down' approach to provide a better understanding of large scale Mesolithic adaptations. The evidence available for an alternative 'bottom up' approach, based on subsistence resources, is addressed in chapter three.

The first 'port of call' in any interpretation of Mesolithic activities is the archaeological record, the 'top down' approach as it is dubbed in chapter one. The archaeological record for the Mesolithic in northern England is however somewhat limited, not in the volume of evidence, there are over 1,900 sites, but in the quality of what is preserved. Almost all the evidence which we have consists solely of 'flint scatters', assemblages of stone tools and debitage, perhaps associated with hearths, or even with luck other features such as post-holes. Exceptional sites, with organic preservation, do exist (Star Carr, Clark 1954, being one particular case) but these are extremely rare. The distribution of Mesolithic sites, and differences in the composition of artefacts between sites, could be a major source of evidence for Mesolithic adaptations. However, these distributions are clearly complex and can not 'speak for themselves'. Even determining what characteristics of these flint scatters could be used to define as a 'site' is difficult (Schiffer 1987; Haselgrove, Millet and Smith 1985; Schofield 1991; Dunnell 1992; Spikins 1995c). Our present understanding of what patterns in the archaeological record say about Mesolithic activities has been built up over many decades. One of the main inspirations being clear patterns observed in the way in which known hunter-gatherers discard artefacts within a seasonal settlement system (Binford 1978; 1980; 1983; Thomas 1981; Wandsnider 1992). As a result, several key ideas which structure our understanding of the archaeological record at a number of scales have their roots in ethnographically derived concepts (discussed in detail in chapter four), from distinctions between 'summer' and 'winter' sites, to ideas of contrasting upland and lowlands patterns of activities for example.

Figure 2.1

Scales of Spatial Patterning.

VARIATIONS OVER TIME Taking first the patterning in sites over time, and how this patterning is commonly interpreted. The clearest element of this patterning is an increase in the numbers of archaeological sites (as discussed in chapter one). Recorded increases in site numbers in Britain appear to be clear-cut and fit in with marked increases in Mesolithic sites recorded in the rest of Europe. Newell (1973: 408) for example documents increases in the numbers of sites of successive periods in the Northwest European Plain, with a threefold increase in the Late Mesolithic.

Patterning in the archaeological evidence can be crudely divided into patterns of change through time, and those of variations in space (the distribution of sites). Temporal changes are coarse-grained, since the typological phases of the Mesolithic are broadly defined and only a limited numbers of sites have been dated by absolute methods. Temporal changes in the numbers of Mesolithic sites can potentially provide evidence for changes in populations, while changes in the characteristics of sites may relate to other changes in adaptations. Interpretations of distributions in space are rather more fine-grained, and can be coarsely approximated at three different scales (figure 2.1). These three different spatial scales of site patterning are illustrated here by 'focusing in' from the scale of the whole of northern England, to that of the Pennines, to that of one area of moorland, Marsden moor. To provide the best potential for interpreting the evidence at each scale, the region with the highest density of sites is selected at each successive stage of the 'focusing in' process. At the large scale, distributions of sites may potentially provide evidence for past Mesolithic territories or the limits of 'social group', at the medium scale, for settlement patterns or the 'seasonal rounds' of groups, and at the small scale for activities in a local landscape. A few interpretations even involve considering changes in the characteristics of sites over both time and space. The interpretation of patterning and the potential effect of different biases on interpretations will be considered in turn.

The apparent evidence for increases in the numbers of archaeological sites through time comes from two sources changes in the numbers of sites which can be grouped into phases according to artefact types (typology), and those which have been given an absolute date using radio-carbon dating techniques. Sites dated by typology in Britain are basically assigned to one of two phases, the 'Early' Mesolithic or the 'Late' Mesolithic. Assemblages dating from the Late Mesolithic are distinctive in containing much smaller microliths than those of the Early Mesolithic (with the exception of the Irish 'Lamian', Woodman (1978a; 1978b)). Other distinctions in raw material use and in other characteristic types of artefacts between the two phases, such as scrapers or blades, often vary regionally, and are discussed in detail by both Jacobi (1976) and Myers (1986). A number of authors, notably, Jacobi (1976), Morrison (1980: 136) and Myers (1986) note clear increases in the numbers of sites which can be typologically assigned to the Late Mesolithic in Britain, over those assigned to the Early. In fact, Jacobi (1976) notes a six-fold increase in Late over Early Mesolithic sites, from a database of 108 sites.

B

The alternative source of evidence for changes in the numbers of sites is the record of radio-carbon dated sites. Smith (1992b) refers to the numbers of radio-carbon dated sites in Britain, and also demonstrates clear increases in sites dated by this method (figure 2.2). Similar increases are also seen within northern England. numbers of sites

40-

13

12

11

10

9

8

7

6

millennia before present

Figure 2.2

Radio-carbon dated sites in the British Isles (after Smith 1992).

The increase in Mesolithic sites through time is even clear at much smaller scales. Differences in the numbers of sites from each phase can be seen in the area studied in detail at the 'medium' scale, the Central and South Pennines 1 (figure 2.1). Although of the 335 sites in this database only 81 could be clearly distinguished as either Early or Late (because of the paucity of documentary evidence from collected assemblages), of these, 23 sites were dated to the Early and 58 to the Late Mesolithic (with 4 having finds from both periods). Clear increases in the numbers of sites in predominantly inland environments are particularly significant in that they provide some initial hope for using gradual population increase as a defining feature of the inland Mesolithic occupation, as well as that for coastal 'complex' societies, an issue raised in chapter one.

VARIATIONS IN SPACE LARGE SCALE DISTRIBUTIONS

At the 'large' scale, that of the whole of northern England, most researchers have been interested in determining first the edges of territories of distinct human populations, and secondly, the patterns of distinct settlement systems, or seasonal rounds. Considering the distribution of Mesolithic sites over space, the 'large scale' patterning of site locations alone has received only limited interest. This may be because it is difficult or time-consuming to collect together the large volume of available material for British Mesolithic sites into a database. No database of Mesolithic sites for the whole of Scotland exists for example. Alternatively it may be because it is difficult to relate the distribution of sites alone to these types of questions. In contrast, the distribution of the apparently 'stylistic' elements of a limited set of sites which have been analysed in detail have been subject to much closer attention and certainly appear to provide important evidence for both territories and for settlement systems.

The Sites Alone The main source of evidence for patterns in the distribution of sites alone, at the large scale of northern England (and indeed for all of England and Wales) has been derived from Wymer's (1977) gazetteer of Mesolithic sites. This gazetteer includes several thousand recorded 'sites' (or unique find spots of Mesolithic material), the northern England component of which (1,987 sites) is illustrated in figure 2.3 (data from Castleford 1987, after Wymer 1977). Although the wealth of locational evidence presented in this gazetteer is remarkable, few direct interpretations of this patterning have been put forward. Using this distribution (mapped according to a 10km2 grid), Smith and Openshaw (1990: 21) have nonetheless noted potential differences in population density across England and Wales as a whole. They tentatively define 'two broadly defined demographic provinces; one in the south and the other in the north and east, with a thinly occupied central zane in between'. Much of northern England lies in the north of these zones, with the Midlands forming an apparently thinly occupied zone between the two regions.

Assemblage Characteristics

1

The site locations and characteristics (Appendix A) are derived from the Sites and Monuments Record (SMR) housed at West Yorkshire Archaeology Service, courtesy of Bob Yarwood and Jenny Marriott, and from the SMR of the Greater Manchester Archaeological Unit, courtesy of Norman Redhead. South Yorkshire SMR and Derbyshire SMR (Sue Whitely and Andy Myers respectively) also contributed data, although since the characteristics of sites were restricted to general terms, only the locational data was used here (as illustrated in figure 2.6).

More interesting patterns emerge when the distribution of assemblage characteristics is viewed at a national scale. Jacobi (1976; 1979: 73) analysed a selection of 54 artefact assemblages from across England. He compared the different types of microliths in each of the assemblages (using cluster analysis) to see which assemblages were more closely related to each other, or exhibited what could be called a common 'style'. He suggested that the distribution of these different types of flint 'industries' at the national scale may relate to distinct 'social territories' in the Late Mesolithic (figure 2.4). Much of the variability which he notes however relates to differences within southern England, with only two of these 'social territories' apparent in northern England. Specifically,

0 •

10

•

100

•

1000

-

100 -

1000 or

more

.. . ...

.... Figure 2.3

Some changes in the character of sites through time are somewhat harder to interpret. One of the most obvious, and not necessarily easily explicable pattern, is that there is a sharp distinction between Early and Late Mesolithic sites in northern England (with no transitional industries being recovered). Transitional industries, such as 'Horsham' industries, do appear to exist in the south of England (Jacobi 1976). Myers (1986; 1989) also notes that the character of upland sites also appears to be distinctly different after the transition. Late Mesolithic sites apparently appear to be smaller and more widely distributed across the landscape than Early Mesolithic sites (although Myers doesn't quantify these distinctions). Myers links these changes to changes in strategies used to hunt red deer in the uplands, proposing a change across the Early to Late Mesolithic transition

10

Mesolithic sites in northern England.

he suggests that northern straight-backed bladelet type assemblages or March Hill industries dominate all of eastern northern England, except for an area of Midlands/East Anglian type assemblages in the south-east of the region (south of the Humber). Jacobi did not classify the west of the Pennines into any typical assemblage type or social territory.

f)

Combining an analysis of lithic assemblages and how these change through time with the distribution of sites also reveals some interesting patterns which appear to be related to changes in social territories. Both Jacobi (1976; 1978) and Myers (1986) have analysed the raw material of Early and Late Mesolithic sites in detail. They note that there is a reduction in the distance over which raw materials are transported from Early to the Late Mesolithic, particularly in the Pennines. In fact, flint raw materials used on Pennine sites in the Early Mesolithic are derived from the Lincolnshire and Yorkshire Wolds (with Early Mesolithic assemblages largely containing over 90% white 'Wolds flint' and in many cases 99% of this material - Jacobi 1976, III. 21; Myers 1986: 311: table 9). By the Late Mesolithic however, raw materials came almost exclusively from local sources. Similar patterns of changes have also been noted in the rest of Britain (Care 1982) and in the rest of Europe (Price 1983; Gendel 1984; Verhardt 1990; Vang Petersen 1984) and are interpreted as relating to a reduction in the size of territories exploited by hunter-gatherer groups, in line with increases in population (as described above), thus providing further support for this concept.

lilJ

.,

{!jl

D

'23s

◊

◊

◊.

◊

t> ◊

♦ ◊

Figure 2.4

rn

Potential 'social territories' according to assemblage types in England (after Jacobi 1979: 68). Different assemblage types are denoted by different symbols, only the in-filled symbols were incorporated into the cluster analysis, the open symbols are additional assemblages.

from an 'intercept' hunting of migrating herds of deer to an 'encounter' hunting of individual animals. In certain situations there appears to be sufficient evidence to link sites with similar 'stylistic' characteristics into a defined settlement system. Both the distribution of raw materials noted by Jacobi (1976; 1979: 73) and the distribution of sites provide a further source of information. Clark (1972) interpreted Star Carr as a lowland winter base camp, contrasting with upland summer hunting sites. Jacobi (1973: 244; 1978: 304) built on this basic settlement pattern, and on Early Mesolithic raw material movements (noted above) to interpret the existence of concentrations of sites in the Lincolnshire and Yorkshire Wolds, 'complementary' to the cluster of sites in the Pennines, (figure 2.5) as winter camps. He argued that groups who occupied the Pennines in the summer would have over-wintered at the Lincolnshire Edge. REGIONAL SCALE DISTRIBUTIONS

'

t

Distinct contrasts between upland and lowland sites become apparent at the regional scale. In terms of the distributions of sites alone, there appears to be a physical separation between sites in the two zones, with few sites in 'intermediate' locations. Thus, in the Central Pennines, upland sites are almost exclusively above about 350m OD (that is, above present sea level) with lowland sites below 100m OD. It is also noticeable that most upland sites occur within a restricted band of elevations, with upland sites being rare above 450m. The distribution of sites does not appear to relate to the most obvious bias, patterns of upland peat erosion, as erosion tends to be most severe at higher elevations. Jacobi, Tallis and Mellars (1976: 308) comment that 'at the highest altitudes where peat erosion is most severe few mesolithic sites have been recorded'. Indeed, a map of the most severe peat erosion (shown for the southern Pennines after Phillips, Yalden and Tallis. 1981) and known Mesolithic sites (from the database compiled here) clearly illustrates that the 'band' of Mesolithic sites is quite distinct from the area of severe erosion, figure 2.6 (the area illustrated is approximately 60km by 60km). An important clue to the distinctions between upland and lowland sites has come from comparisons of selected artefact assemblages. Mellars (1976) compared upland sites (from across England and Wales), characterised by assemblages dominated by microliths and by small site dimensions, with larger lowland sites, of which there were two types, commonly dominated by scrapers. Microliths are usually interpreted as the 'barbs' for arrows used in hunting, and scrapers as used in 'domestic' activity. On the basis of models from ethnographic sources (discussed in chapter four) Mellars interpreted the small upland sites as 'hunting camps' likely to be occupied in the summer, and lowland sites as more likely to be longer term occupation, or 'base' camps, occupied in winter. The distributions of sites at a regional scale, noted above, confirmed the idea of distinct

? Early Sites

Barbed Plont Groove & Spllnter

■

Bone Pick

Distribution of Early Mesolithic sites in northern England: Br - Brigham; D - Deepcar; DH - Dowel Hall Cave; Ho - Homsea; HP - Holme Pierrepoint; Ma - Manton Warren; MC - Misterton Carr; MGP - Mother Grundy's Parlour; Ra - Radcliffe Ees; R - Risby Warren; S -Skipsea; SC- Star Carr; U - Uhome; WH-WarcockHill; Wi- Willoughton Site A; WN - Welton Mill Minor Rock-shelter.

Figure 2.5

At the regional scale, interpretations have focused on identifying the types of activities which occur in different environments, as well as providing specific explanations for differing distinct local patterns.

O T A

Concentrations of Early Mesolithic sites in the Pennines and Yorkshire Wolds (after Jacobi 1978: 304).

upland hunting grounds (Jacobi 1978), above which there was little occupation. Following Clark (1972) it was widely suggested that the use of the uplands was probably restricted to the summer months (Clark 1972; Mellars 1976), when hunting deer. Contrasting local patterns are also a component of observations at the regional scale. Certain local areas have very high relative densities of sites, which have been interpreted by early researchers to be a result of movements of populations through key areas. However, complex patterning in the assemblage characteristics of sites at the regional scale has also been determined, and which is much more difficult to interpret. Several researchers have noted that upland sites can be markedly varied in their assemblage characteristics, and moreover, that different 'styles' of assemblage can overlap within the same region. The most obvious concentration of sites is that at Marsden moor, the area chosen as the local example here. Marsden moor is in fact the location of the highest density of recorded Mesolithic find locations in England and Wales. Buckley (unpubl.) interpreted these high densities in terms of the topography, although later authors have not determined any obvious explanation (Stonehouse 1990). Since this region is at the narrowest part of the Pennine chain, Buckley supposed that Mesolithic populations moving north-south along the Pennines were constricted into a smaller area at this point and left behind higher densities of artefacts. A similar explanation for relatively high concentrations of find locations has been proposed in other areas, such as for Rombalds Moor (see figure 2.6) by Cowling (1946). He suggested that Rombalds Moor may have been a key passing place across the Pennines throughout Prehistory, and that this location may explain the concentration of Mesolithic (and Neolithic) sites. Differences in the artefactual composition can add a great deal to an understanding of regional processes, although detailed analyses require a long-term commitment to

.

,,-..,,

\:). '\·.:~;,.

1 Okm Figure 2.6

Sites in the southern Pennines and areas of most severe peat erosion.

analysing many assemblages. Fortunately, the sites within the Pennine area discussed here formed a major component of a number of national scale analyses (noted above), which have revealed distinct patterns at the regional scale. However, whilst contrasts in artefact assemblages appeared to be explainable at a national scale, potential explanations for regional patterns are much more difficult to identify. In terms of the distribution of Early Mesolithic assemblages, one of the most interesting analyses including Pennine evidence is Jacobi's statistical analysis of a series of 108 assemblages from across England (Jacobi 1976; Switsur and Jacobi 1975; 1979) (noted above in reference to large scale social territories). Jacobi identified two distinct types of assemblages in the Early Mesolithic, which have different microlithic forms and raw materials. In the Pennines these two distinct styles are found in close proximity, having even been recovered from the same hillside (the sites of Warcock Hill South and Warcock Hill North). The former site, W arcock Hill South, appeared to be related to Star Carr in the Vale of Pickering, on the basis of stylistic traits and raw material sources, (Radley and Mellars 1964: 21), whilst the assemblage from Warcock Hill North has more local affinities. The potential explanations for these two site types are varied, and could include not only functional differences in the two types of sites, but occupation by different huntergather groups or even a chronological distinction between the two sites.

In the Late Mesolithic, the Central Pennines are again the focus of attention in Jacobi's study (Jacobi 1976; Switsur and Jacobi 1975; 1979) with two distinct assemblage types recovered here. As well as the most common 'March Hill industries' (dominated by 'scalene triangle' microliths) recovered from this area, there are also distinctive 'rod microlith dominated assemblages' (Jacobi 1976; Switsur and Jacobi 1975; 1979). These sites are only found in the Central Pennines and in the North York Moors. They tend to only occur at high elevations, (not in the lowlands), often being re-occupied, with the only raw material used being flint rather than the range of flint and cherts common in Pennine assemblages, (Jacobi and Switsur 1975; 1979). A recently excavated 'rod' microlith dominated site excavated at March Hill Top demonstrated all of these characteristics, and also contained a small hearth, with at least two phases of use, (Conneller 1995; Spikins, Ayestaran and Conneller 1995; Spikins 1995b). Again the evidence from these two types of assemblages may relate to functional or even social group contrasts, although chronological distinctions are a distinct possibility given the very late dates for these types of sites (Jacobi and Switsur 1975; 1979), especially since the dates from March Hill Top cluster at around 5,200bp (Spikins, Bayliss and Bronk-Ramsey, in prep). Within the Pennines themselves, Marsden moor has again been the focus of attention in studies of distinct assemblage 'styles' as well as in reference to site densities. This area is

one of only two local landscapes where very rare 'pear' microlith type sites have been recovered, (Jacobi 1976; Stonehouse 1987; 1990), the only other area being in the Lincolnshire Wolds (Jacobi 1976). The distributions of sites within Marsden moor are considered in more detail below.

A

Large Assemblage

•

Small Assemblage

.....

LOCAL SCALE DISTRIBUTIONS

At the local scale, it is general patterns, apparently common to all upland distributions which have attracted the most attention. These general patterns are interpreted as relating, not only to the types of activities occurring in the uplands, but also to broader issues such as the long-term continuity of these activities through time. The most notable element of small scale patterning is that most upland sites tend to be found at a specific elevation, at points from which there is a high visibility of the surrounding area, as well as tending to be found on southfacing slopes and at river heads, as clearly illustrated by the distribution of the large number of flint assemblages found on Marsden moor (figure 2.7), and the distribution of typologically dated sites in this area (figure 2.8).

• ••···Dean

......... ..... ...... .._

'

····45om.... •········

...

.

Figure 2.7

Mesolithic sites on Stonehouse 1990).

Marsden

moor

(data

from

Figure 2.8

Early and Late Mesolithic sites on Marsden Moor.

Buckley (1924) was one of the first people to notice this patterning. He commented that In this district (around Marsden), the sites chosen as workshops or camping grounds were comparatively small and well defined, situated on the tops and upper slopes of hills and ridges, and at least 1,250ft (381m) above sea level.

Radley and Marshall (1963: 96) later also commented that in the Central Pennines 'Mesolithic sites prefer the J,250J,500ft zane [381-458m] on east to south-east facing slopes', and Barnes (1982: 25) interpreted 'sunny slopes between 1,200 and 1,500 feet [366-458m] [as] being favoured localities overlooking spring heads'. The most popular interpretation of these characteristics is that they are the preferred locations of Mesolithic populations, being probably the best 'lookout' sites for hunting groups, who may have been waiting for passing red deer. Thus, in discussing the use of the uplands by hunting groups watching for deer, Jacobi supports his argument by noting that Mesolithic sites are 'clustered on certain ridges, hills, 'edges', valley heads or eminences, each one controlling the maximum possible view ... many of the sites overlook natural basins ... situated to take into view the largest area possible' Jacobi(1978: 325)

Most recently, Simmons (1996: 33-34, cited at the beginning of this chapter) has drawn on this element of patterning, again using the Marsden moor example. He related the distribution of sites to a long term continuity of use of the uplands, both directly for the hunting of red deer, and indirectly through clearance of vegetation to increase the quality of browse for these animals. He interpreted upland clearance phases as intimately tied to the same types of location in which clusters of Mesolithic sites are recovered.

Also of importance is the continuity of upland activities across long time periods, a common and important theme tending to support ideas of a continuity and a stability of settlement. Common site 'preferences' link not only the Early and Late Mesolithic (Myers 1986), in Marsden (figure 2.8), the wider area of the Pennines, as well as elsewhere (Barton et al. 1995), but also the Mesolithic and Neolithic. Tilley (1994) strongly argues for continuity in the use of upland landscapes. Drawing on evidence for the same topographic preferences in south-west Wales, he argues for a more widespread continuity in the symbolic importance of particular 'locales' throughout the Mesolithic and Neolithic. He remarks on this 'continuity in the choice of locales and the exploitation and use of particular areas of the landscape.' (Tilley 1994: 145), also noting that for Mesolithic populations 'evidence that these populations had a specific affinity with particular... locales and areas is overwhelming' (Tilley 1994: 84).

Spikins (1993; 1995c) has attempted to quantify the regularities identified in site distributions in the Pennines. This work need not be presented here, but in simple terms involved comparing find locations (326 'sites'), and an equal Elevation (sites) 1 0

F8 r

the most heat from the sun, and thus being the driest and warmest spots. Kvamme and Jochim (1985) also noticed similar patterning to that identified above when statistically analysing the locations of upland Mesolithic sites in Germany. They also interpreted the patterning shown as difficult to relate to known biases, and therefore most likely to be a result of 'real' preferences exerted by Mesolithic populations.

e

It is also worth noting that evidence for specific activities

q 6 u e 4 n C 2 y 180.0 260.0 340.0 420.0 500.0

Elevation (m) Elevation (non-sites) 1 0

F 8

r e

q u

6

e 4

n C

60.0

140.0 220.0 300.0 380.0 460.0

Elevation (m)

Figure 2.9

Elevation of sites and non-sites in the Pennines (after Spikins 1995c: 95).

random sample of points ('non-sites'), with coverages (maps) of elevation, slope, aspect and distance to minor and major streams (generated from topographic data at 50m resolution). The relationship between the types of landscape in which sites were recovered, and the more general characteristics of the wider landscape were thus able to be analysed statistically. Several elements of patterning were identified as statistically significant (using logistic regression techniques) however Spikins (1993: 1995c) noted that some of the patterning is likely to be a product of biases in the visibility of artefacts or the actions of collectors. There were however several key elements of patterning which were difficult to explain by reference to these factors. Spikins concluded that the elevation of sites, and the aspect (the direction in which the sites face) related to genuine preferences exerted by Mesolithic populations. A plot of the distribution of elevation values for 'sites' and 'non-sites' for example demonstrated that the find locations showed a restricted distribution across the possible elevations, a characteristic also noted by Buckley (1924) Mellars (1986) and Stonehouse (1990), (see figure 2.9). Equally, the distribution of recorded site values for aspect shows a tendency for find locations to be preferentially located on south-east facing slopes, which was interpreted as relating to the locations most attractive for hunter-gatherers, receiving

which were carried out in the uplands is also recovered from within sites. In fact, a further scale of analysis within Marsden moor itself could be defined, particularly at a series of recently excavated sites on March Hill and Lominot. Here the distribution of artefacts has been analysed in detail and related to several separate sequences of knapping activities around central hearths (Spikins 1994; 1995c; 1996; Spikins, Ayestaran and Conneller 1995). The intra-site scale is not included here in more detail as the interpretations at this scale rarely relate to the large-scale processes of adaptation which are addressed here. Moreover, distributions within sites are subject to different types of biases than those addressed here (Spikins Ayestaran and Conneller 1995, for example, discussed recovery biases on excavated sites). A detailed discussion of evidence at this scale, and interpretations of this evidence can be found elsewhere (Spikins 1994; 1995c; 1996; Spikins, Ayestaran and Conneller 1995). A final point to be made before proceeding is that the distribution of sites within local, regional or national landscapes are not a component of all interpretations of Mesolithic adaptations. Some studies concentrate on the evidence from single sites for example, such as re-fitting studies (Barton 1992), or microwear analysis (Dumont 1988), and may not draw on wider scale distribution patterns. However, for the majority of interpretations of Mesolithic adaptations, and the broad characteristics of the Mesolithic occupation of northern England, the distribution of sites is a key element of interpretations. It is clear from the above discussion that the distribution of

sites is a major component of interpretations ranging from changes in population numbers to large scale differences in population densities, the organisation of settlement systems, or activities at a local level, and a continuity of these activities through time. The interpretations put forward about these activities appear to be quite reasonable and logical. However, any biases which affect these distributions will have far-reaching effects on our interpretations. Unfortunately, many site distributions are substantially biased. Biases affecting the temporal or spatial distributions of sites at different scales not only affect the validity of interpretations based on these distributions alone, but can also affect interpretations based on patterns in the characteristics of sites.

THE EFFECTS OF BIAS

120 100

Biases have affected the distributions which form the basis for interpretations of key characteristics of the Mesolithic occupation of northern England in a number of different ways. They can affect the recovery and identification of sites, and also even their interpretation. Similar broad types of biases affect both the temporal and spatial distribution of sites, however certain specific types of biases are unique to either situation and thus temporal and spatial biases are considered in tum.

80 60

40

20 0 0

0

0

"

C)J

0 0

'I'

;;;

0 0

'I'

0

0 0 0

0 0 0

0

0 0

"'

'I'

0 0 0

"'A

assemblage size

THE TEMPORAL PATTERNING OF SITES

It is often assumed that known dated sites are a direct reflection of 'what is out there' in the archaeological record (or what was deposited in the past), although in reality this is far from being the case. Different biases affect the recovery and identification of sites in different ways. The effects of different topographic and geological conditions, and different human factors such as the intensity of collection are considered in detail in the following section. These may affect how representative dated sites are of population if, for example, settlement systems changed through time and locations where sites are 'preferentially recorded' today were used differently in the past. What is considered here however are those biases which directly affect the relative recovery and identification of sites dated to different periods. The most obvious bias to have affected the recovery of sites dated to different periods is the influence of the stratigraphic location of artefacts. On almost all excavated sites the Early Mesolithic artefacts are recovered from levels beneath the Late. This means that in very simple terms, Early Mesolithic material has a lower chance of being recovered as it is deeper in the sediment (by about 6cm at March Hill, Spikins 1995b, Spikins, Ayestaran and Conneller 1995). This stratigraphic location of different sites might appear to be a minor affect, however it does appear to have a real influence on the numbers of recorded sites by acting against the recovery of earlier artefacts - excavations carried out as part of the West Yorkshire Mesolithic Project clearly demonstrated that many flint collectors failed to 'dig deep enough' to recover all of Early Mesolithic scatters (Spikins 1994). Once recovered, sites which are assigned to different periods on the basis of typology may be subject to different probabilities of being correctly identified. Differences in the use and form of the main diagnostic element of Mesolithic assemblages - microliths - may markedly affect how sites of either period are recognised. Myers (1986; 1989) notes for example that there is a higher ratio of microliths to other artefact types recovered on Late rather than Early Mesolithic sites (Myers 1986: 235: table 5). The chance of recovering a diagnostic artefact (a microlith) is thus greater in any assemblage of Late Mesolithic artefacts. The fewer microliths likely to be found on Early Mesolithic sites effectively acts against these sites being identified in comparison to Late Mesolithic sites. Since most recorded 'sites' are often a collection of only a few artefacts and can't

Figure 2.10

Numbers of artefacts on Mesolithic sites in the Pennines (from a database of recorded sites, derived from Sites and Monuments Records for West Yorkshire and Greater Manchester).

be dated (figure 2.10), we would expect, other things being equal, that many more Early rather than Late sites remain unidentified. There are also factors which act against the identification or recovery of Late Mesolithic sites. The tiny dimensions of Late Mesolithic microliths make them more difficult to see than larger Early Mesolithic forms, and the typically dark brown/grey flint or black chert of which they are produced makes them more difficult to distinguish from the peat/soil substrate than Early Mesolithic microliths, which are often made of white flint. Even if recorded dates were representative of the 'real' pattern of Mesolithic sites, the interpretation of this record may be biased. For one thing, a change in settlement patterns may mean that a supposedly similar 'site' from one period 'meant' something very different in the next period in terms of population numbers. Another factor to consider is the time at which different regions were colonised. One factor that may be affecting Smith's (1992) increase in site numbers may be that some new areas are colonised within the Mesolithic. Figure 2.11 shows that when dated sites are plotted against Northings, there is distinct differences through time in the area of Britain from which sites are recorded, with most Scottish sites (above about 550000 North) dating to the Late Mesolithic. Even the way in which information is presented can generate biases. One very simple bias which affects common conceptions is that it is misleading to directly compare numbers of sites from the Early to Late Mesolithic to each other. Graphs often compare Early and Late Mesolithic site numbers using equal 'time blocks' (Jacobi 1976 for example). Although the two time periods are often seen as complementary, in fact the Late Mesolithic spans a time period approximately twice as long as the Early. This means that differences between the two periods are easily conceptually inflated. One very different factor uniquely affects the interpretation of radio-carbon dated sites. Radio-carbon dates are not in fact, a 'true' record of past dates. The sequence of radiocarbon dated sites is in reality affected by variation in the atmospheric content of Carbon 14 , which makes a direct linear

11,000

10,000

9,000

8,000

7,000

6,000

5,000 BP

Typology

(Myers 1989, Switsur and Jacobi 1979)

Early Mesolithic Late Mesolithic Later Mesolithic

Radiocarbon dated sites v Northings

(data from Smith 1992)

.--------------------------------------------------,

1000000 900000

•

t-+--1 111 .... 1 •

ll+H·#H i+i ♦

I

i+i

♦

r+--/+i I

Ill I

I ♦

t---+---i

~ I ~+-++--I

♦ -lltlll

1----+----l

' .... i-+--l .--+:p~III

.... ~ t-+--1 ul

t-+-l

I

10000 BP

600000

♦

I 4111 I t-+--1 400000 1----+----l 1--+--i

f-+-1

1----+----l

I ♦

r+-l

I

300000

1----+----l

t-+--1

200000

t---+---l~•~1•111W1 ♦ II I .. I 111• ♦I I t---+---i

I •

H♦♦-t

1----+----l

............... --i---'-------' 9000 BP

8000 BP

7000 BP

100000

i----+-----i 1----+----l

'-----------------------------------------....l. 11000 BP

700000

:~

500000

- •1~•1

~

riri•I"" ~ n• ~ I ~

,■

r+-l

.IJl/:iiaill1-I 1

,■■-

i-----+-----l

1----+----l

t---+---i

"'"::

t-+--1

i----+-----i1~11--+--i

1----+-----l f-41-+-1 t----+----1

,nit.1-

♦II Jt"7 t+-l

II ... II 111.-i

1----+-----l

1----+----l ~t-+--1 1----+----l

•

1•1t~I I

I

111:.1 I H+,+----,i

I .. llftlllllllllll-111I ~

I

800000

t-+--1

I

t-+-lt-+--1

♦

t+-l

6000 BP

0 5000 BP

C14 Date (BP) Figure 2.11

Mainland Britain radio-carbon dates v. Northings, with Northern England typology. a) dated sites

date (the uncalibrated date) only a skewed representation of the 'real' date. Dates can be calibrated using the 'radio-carbon calibration curve' to give a date in real years (uncalibrated dates are commonly given the suffix bp or be and calibrated dates BP or BC) which often gives a very different result from the 'uncalibrated' plot. Thus, a plot of calibrated dates (using CALIB 3.0 - Stuiver and Reimer 1993) (using the central mean where more than one mean date is calculated as the calibrated date) shows a much more complex picture than that of the uncalibrated dates demonstrated in Smith (1992), figure 2.12, both for the British Isles as a whole and for northern England.

15

14

13

12

11

10

8

9

7

6

millennia before present b) dated sites

Although the archaeological evidence for gradual increase in population in the Mesolithic appeared to be clear-cut, a closer consideration of biases has revealed a very different picture. There still appears to be an increase in Mesolithic sites assigned to the Early or the Late Mesolithic (according to Jacobi's analysis), but some doubt has been cast of the reliability of evidence for clear gradual population increase on the basis of a change in the numbers of dated Mesolithic sites in the British Isles. The picture is evidently a complex one and a better understanding of any possible changes in other adaptations (such as changes in the settlement system or the location of sites) may provide a better context for understanding potential changes in population. Before doing this however, the biases acting on the spatial distribution of the sites must be addressed.

14 12 10

8 6

4 2 0

..._ 15

14

13

12

11

10

9

8

7

6

millennia before present Figure 2.12

Calibrated dates a) for the British Mesolithic and b) for northern England (original dates from Smith

1992).

rn

BIASES IN SPATIAL DISTRIBUTIONS

TYPES OF BIAS

Whilst temporal biases may be straightforward to approach, biases in spatial distributions are rather more difficult to define. As with temporal changes, biases can affect the spatial distribution of sites not only through the recovery of artefacts but also their identification and interpretation. Effectively, the sequence of processes which affect the probability of any 'site' being recorded can be divided into two components - the relative visibility of artefacts at the soil surface, and the probability of their subsequent recovery and recording.

Different types of biases operate at different scales, and the relation between different factors of bias are often complex and inter-related. We can however broadly define the types of bias and their effects at the three different scales considered.

Visibility Artefacts are exposed to surface collection and are thus more likely to be recovered where there are natural processes of erosion, ploughing or any other human disturbances which expose artefact levels (such as engineering works or road projects). In contrast, they are very unlikely to be recovered where they are deeply submerged such as where sedimentation rates are very high (such as in alluviated river valley bottoms), or in areas which have been inundated by rising sea-levels.

Recovery Even if artefacts are visible they have a probability of being recovered and recorded, which relates to the number of individuals who pass, collect and record the presence of artefacts. This is essentially related to the ease of access to the location, such as the proximity to roads and footpaths, as well as to the presence of nearby populations within which some people will be interested in looking for Mesolithic artefacts, and moreover prepared to relate their findings to museums. In many cases this latter factor relates to the presence of local archaeological societies and other similar groups.

The Underlying Distribution? The two processes of visibility and recovery are often conceived of as a 'window' of opportunity for collection, underneath which is the underlying artefact distribution, whose potential recovery depends on preservation (in most of Britain flint or chert artefacts are the main surviving materials from the Mesolithic) and which may be subject to distortions due to sediment transport and erosion. The real situation may however be more complex since interpreting this underlying distribution is also far from straightforward, for one thing, the characteristics of the person who finds a 'site' will also influence the identification of sites.

Natural Variations in Soils and Topography The natural variations of the landscape - the different geology, topography and superficial deposits of different regions of northern England - have had the most obvious effect at the large scale. These factors include large scale movements of the land surface and sea levels, as well as general patterns of distinctions between different topographic locations and the influence of regional geology and soils. The simplest explanation for one element of site distributions relates to the effect of isostatic uplift and sea-level change since the last glaciation, considered in chapter one (as discussed by Tooley 1974; 1978; Sherman 1989; Lambeck 1995). The melting of glaciers since the last glaciation caused the volume of water in the oceans to rise and a large part of the evidence for lowland occupation in the Early Mesolithic (about half the total area of the lowland Mesolithic landscape) has been obscured through the effect of rising sea-level. This melting of upland glaciers also removed some of the load on continental areas which subsequently rose, such that the position of the changing coastline is determined by a complex relationship between this 'isostatic recovery', movement of the underlying mantle and sea-level rise. These processes have been modelled in detail by Lambeck (1995). The main axis of present relative uplift (versus down-warping) runs south-west to north-east across the northern part of northern England, with land in the north-west rising and the south-east falling (as illustrated by Goudie and Brunsden 1994: 27, after Sherman 1989). Sites have been submerged by rising relative sea-levels in the south-west and south-east of northern England with coastal sites in the uplifting north-west and north-east remaining visible. Farther north in north-west Scotland, surviving coastal Mesolithic sites provide our main source of evidence for coastal exploitation patterns in Mesolithic Britain (Mellars 1987; Bonsall 1996). The contrast between uplands and lowlands, through the presence of peat in the uplands, has also had a particularly marked effect on recorded site distributions. The formation of peat is affected by both the geology and the topography of the English uplands. The flat pleateau topography of the Pennines is a major factor encouraging peat formation for example (Taylor 1983). Peat formation is a complex process, which on many of the upland areas of northern England was initiated at different times. Human disturbance has had a major role in the development of peat in water-shedding sites, however in water collecting sites it appears that the change to wetter climates during the Holocene was a major factor causing peat accumulation (Simmons 1996: 133). The precise relationship between human causes (such as vegetation clearances) and natural processes (of progressive water-logging and leaching of upland soils throughout the Holocene) on local histories of peat formation is not agreed

upon (Taylor 1983: Simmons 1996). Local variation in the timing of peat formation can in fact be high. At one valley in the North York Moors (North Gill) Simmons (1996: 99) demonstrates that peat formation in the upper section of the valley was initiated during the first half of the ninth millennium bp, in the lower section in the eighth millennium bp with peat spreading to the gentler slopes, and in the middle section during the sixth and fifth millennia bp. Peat in principle covers and preserves Mesolithic artefacts. However peat erosion, from the last century onwards, has been severe enough to 'cut through' both the peat and the old soil surface (the 'palaeosol') in many areas, in which cases artefacts previously preserved at this buried soil surface become exposed. It is the erosion of peat which has greatly contributed to the much higher densities of recorded Mesolithic sites clearly linked to the upland areas of the Pennines and North York Moors (shown in figure 2.3). In fact, although not all uplands are peat covered, and erosion only occurs at specific locations within peat covered area~, the density of known Mesolithic sites above 300m OD 1s three times the density of sites below this elevation 2 •

Figure 2.13

Peat erosion rates are governed by a number of human factors (discussed in more detail below). Although direct human erosion through trampling at popular footpaths is one factor, high levels of atmospheric pollutants which destroy vegetation (especially sphagnum moss), as well as ~he intensity of grazing sheep (which both destroy the vegetat10n cover and physically erode peat itself), have been more influential. Massive upland peat erosion was initiated in the late 19th century largely through a combination of a rise in pollutants and a rise in grazing intensity (Phillips, Yalden and Tallis 1981). The effects of these factors can be considered separately although they are clearly closely related.

pollution (from mining) still have many mosses growing in the peat areas, most probably because they have suffered less from gaseous pollutants (Lee 1981). The co-incidence of the highest density of recorded finds between the key industrial cities (figure 2.13) may be no accident. An increase in pollutants in the middle of the last century also coincided with high populations, a demand for meat and a much increased intensity of upland sheep grazing, especially near major industrial towns. Sheep not only directly eat vegetation, but are also effective at trampling sensitive species, and at actively eroding peat haggs when resting. The numbers of sheep on the moors have been increasing rapidly since the middle of last century (see table 2.1) and continue to create an erosional problem. Grazing pressures of less than 0.6-0.8 sheep/hectare are necessary for heather and bilberry to survive in appreciable amounts (Yalden 1981), however densities in the Peak District in recent years have typically varied from 1.67-4.54 sheep per hectare (ranging from a minimum of 0.54 to a maximum of 8.12), (Phillips 1981) with even higher effective densities where much of the vegetation (such as matt grass) is now unpalatable to sheep (Evans 1992: 55).

Atmospheric pollutants have been a factor of daily life around the heavily industrialised cities at the heartland of the industrial revolution, such as Bradford, Leeds, Manchester and Sheffield, for over a century. The neighbouring uplands have also been affected. Atmospheric pollutants can be either particulate or gaseous. Particulate matter, especially soot, is a major factor influencing rates of peat erosion in uplands in the Central and South Pennines (Johnson and Dunham 1963). Soot, observed in several peat profiles in the South Pennines (Lee 1981), coincides with the disappearance of mosses and peat erosion dated to the middle of the last century (Tallis 1964; 1990). As Nowell (1866) remarked (cited in Lee (1981)), the disappearance of a number of mosses from the Todmorden area correlated with a 'super-abundance of smoke'. Gaseous pollutants have also had an effect in inhibiting vegetation growth and encouraging peat erosion, and again this pollutants tend to be concentrated in the Central and South Pennines. The North Pennines despite lead 2

Main centres of population in northern England.

Almost all upland moorlands have suffered the effects of over-grazing and consequent peat erosion. Young (1986) notes that the main factor influencing recovery in the North Pennine uplands, like those to the south is also peat erosion.

In the database derived from Wymer (1977) considered here, there

were a total of 64 7 find locations above the 300m contour and 1,340 below it. The total area of the former was 6, 179km2 and the latter 41,140km 2 (as recorded in Arclnfo from the digitised topography), giving a density of 0.0325 finds per km 2 in the lowlands and 0.090 finds per km 2 in the uplands. The figures are meant only to illustrate the relative densities, not as precise measures of finds densities in the two environments.

1805 (Luccock 1805)

1874 (MAFF 1874)

1973 (MAFF 1973)

North Riding

365.3

731.5

880.8

West Riding

383.1

770.6

896.0

Table 2.1

rn

Numbers (thousands) of sheep in the North and West Riding of Yorkshire (after Evans 1992).

historian]would surely tum in his grave could he but see, the terrible way in which March Hill has been cut to pieces.

Numbers of recorded sites

30 25

The distribution of peat and peat erosion are not the 'whole story' of site distributions. Most other effects of natural physical variations are however much more specific to particular regions. One that is particularly important in the context of interpretations of settlement systems (Jacobi 1973; 1978) is the marked contrasts in recovery conditions from the Pennines to the Yorkshire and Lincolnshire Wolds, across the floodplain of the Humber estuary.

20 15 10 5 0

A consideration of the deposits (figure 2.15) in this area suggests that a number of factors may be 'inflating' the densities of sites at either end of Jacobi's (1973; 1978) settlement system, whilst at the same time the densities of sites in the 'middle' of the proposed settlement system may be artificially reduced.

decade of recovery ( recorded for 103 sites) Figure 2.14

Numbers of recorded sites by decade in the Central Pennines.

In terms of the two ends of the 'settlement system' it is already clear, as noted above, that sites in the peat uplands are exposed by erosion. However, two factors may also be encouraging the recovery of sites in the Lincolnshire Wolds. Both these factors relate to the local predominance of calcareous bedrock - in the case of the Wolds, soft chalk soils. First, these soils are particularly prone to erosion. Evans (1977: 58) for example demonstrates values of 90200mm per year of total lowering through recent erosion of chalk substrates. Secondly, the chalk soils are a major source of flint, and regions closer to a flint supply might be expected to yield a greater density of artefacts. In simple terms, where flint is in short supply, artefacts are re-used more intensively (so fewer enter the archaeological record) are smaller (and more difficult to find) and also other material may be used rather than flint where this is possible.

alluvium calcareous bedrock peat uplands

Figure 2.15

Deposits affecting the visibility of sites in south-east northern England.

The initiation and continuation of flint collection activities, the recovery of artefacts, made visible through erosion, although linked to the antiquarian tradition, appears to be closely related to the subsequent initiation and severity of peat erosion. Dates of finds collection from the Central Pennines thus begin when peat erosion is first initiated and continue to the present (figure 2.14) (the effect of the second world war in removing flint collectors 'from the scene' is also clear in this figure). The area chosen for the local scale study, Marsden moor, was once severely eroded although it now suffers from only limited erosion (thanks to reduced sheep numbers under the ownership of the National Trust). Photographs from the 1940s to 1960s document a high level of erosion, and a level of collection which followed suite (aided by the ease of access to this site, a factor discussed below). A local collector, J. L. Turner (1964) noted in the 1960s March Hill, the mecca of all true flint addicts ... this place is in absolute turmoil being slashed, hacked and tom to pieces in a most sacreligious way. Ammon Wrigley [a local

The effect of flint availability in the past on present finds densities can be illustrated by recent regional surveys of lithic scatters. One example of a fieldwalking survey carried out in an area which had no local flint sources was that in the Tyne-Solway valley in the north-east of England, conducted by Tolan-Smith (1996). An example of a survey carried out in a similar environment where flint sources are local could be a survey in Hampshire carried out by Sherman (1985: 50). It is very difficult to compare data from fieldwalking projects, since recovery rates depend on many different factors (such as the depth of finds, type of ploughing experience of collectors). Comparisons may not be accurate, however the contrasts between the two surveys are nonetheless remarkable. In the former, collected finds densities are about 0.97 items per 1000m2 (Tolan-Smith 1996: 9), in the later on the other hand, finds densities reached an average of 79.5 artefacts per 1000m2 (Shennan 1985: 50). Though incidental to the main discussion here, this lack of local sources in this north-east region may be one factor effecting the relatively paucity of sites in these areas (clear from figure 2.3), and at least making it very difficult to separate potential differences in past population densities from biasing factors. One major factor acts against the recovery of sites in the intermediate zone of Jochim's (1973; 1978) settlement

rn

Human Factors

system. This is the large deposits of alluvium which have started to accumulate since the Mesolithic (Jones 1993: 257). The accumulation of alluvium would effectively obscure sites beneath a dense layer of silt, preserving sites at a deeper level than ploughing or construction activities typically disturb.

Human factors influencing site distributions include broad biases resulting from patterns of human exploitation of natural landscapes, different land use practices and variable population densities, as well as more 'individual' human factors such as the influence of particular individuals on the record of sites.

Essentially, whilst sites in the Pennines and in the Lincolnshire Wolds in the Early Mesolithic may be linked by raw material and artefact similarities, there is little real evidence that they formed either 'end' of a settlement system. At the least, many other sites which have yet to be recovered may also have been part of a wider regional pattern.

Different land-use practices can have a marked effect on the visibility of sites. The influence of grazing pressures causing erosion within peat uplands has already been noted. In the lowlands, the main factor affecting finds recovery is ploughing. Ploughing on arable farms only reveals only a limited number of artefacts at the surface which are only occasionally recovered and reported. Schofield (1991: 101) carried out experiments which suggest that the maximum recovery of surface assemblages is only 3.5% of the actual ploughzone assemblage (this figure is for flakes and tools, a figure of 0.5% being taken as a maximum for cores), these precise numbers of artefacts of course depending on the depth of artefacts, and of ploughing as well as other factors such as the substrate. Nonetheless, though only an element of total artefacts are revealed, the effects of ploughing in heightening the visibility of artefacts is clear. For the sites in the Central Pennines (listed in Appendix A), where the method of discovery was stated (only 76 sites), 61 were recovered due to erosion (almost exclusively of upland peat as previously disussed), 10 from ploughed fields, and only five from other causes - 2 from quarrying/gravel works and 3 from forestry work. The effects of ploughing are also one of the main influences on the recovery of sites in Weardale (Young 1986; 1987).

Different geology and soil types also affect other regional patterns. Myers (pers. comm.) notes for example a relative increase in the densities of sites on limestone areas in Derbyshire, where thin soils make the chances of recovery easier. The same process has also been noted at Malham Tam (Donahue 1996), particularly where finds are easily brought to the surface by mole action. The types of flint that are found in different deposits, different chalk regions, or different local river gravels, may even have had an effect on the distribution of different categories of sites. The nature of the raw material used when knapping flint tools can have a major effect on the end product - small flint nodules restrict the user to the production of small flakes and blades for example, while certain types of flint and chert can be used to make finer blades. Although in Europe evidence has been found for stylistic traits that cannot be linked to raw material constraints (the right and left lateralisation of microliths, Gendel 1984; 1987), it is certainly possible for the style zones identified by Jacobi (1979) to be at least influenced by raw material source areas. It may be 'suspicious' that the Midlands/East Anglian type assemblages relate to the area of use of Wolds flint in the Late Mesolithic, and the northern straight-backed bladelet type assemblages or March Hill industries to the use area of local valley gravel flints. How accurately the 'style' of artefacts types represents past social groups is another issue (Hodder 1982). Further work on these assemblages would however be needed to test this suspicion.

Whilst differing intensities of upland sheep grazing, and lowland arable farming, have a local effect, they can however also have a wider effect at the regional scale, and an effect which may be relevant to the issue of distinct upland and lowland patterns of activity. Whilst it is often considered that, in simple terms, upland peat areas are the preserve of hill sheep farming, with lowland areas being the preserve of both arable and livestock farms, this is not strictly the case. Sheep farming is in fact the dominant land-use not only on peat uplands, but also on the lower flanks of these uplands, down to flat lowland areas (Evans 1992). Essentially whilst artefacts in peat covered areas may be relatively more 'visible' than in the intermediate zone due to peat erosion, those in lowland areas may be more 'visible' than in the intermediate zone due to arable farming and ploughing. Evans (1992: 56) for example states that 'between the slopes [of peat covered uplands] which are susceptible to overgrazing and the arable fields is a zane of (now improved) grassland where erosion is rare'. In effect, it is possible that the supposed evidence for a distinct division between upland and lowland sites may be more a factor of modem land-use practices than any distinction in the past intensity of activities.

Unique geology and soil conditions can have effects at the local scale. One factor which may be influencing the differences in local densities of recorded sites along the Northumberland and Durham coast in the north-east for example, is that sand dunes north of the river Tees (shown in Goudie and Brunsden 1994: 45) would both offer more opportunities for access to the coast in the past, and for the exposure of artefacts in the present than the predominance of cliffs to the south of the Tees (Goudie and Brunsden 1994: 50). As noted above, many distributions are the result of complex interplay of different biasing factors. Natural landscape variations are found at different scales, with these processes combing with 'human' factors to affect site recovery, whilst other 'human' factors to affect the identification of sites. In particular 'human' factors of bias are evident at all scales, but the effects are clearest at the medium scale.

Differences in land-use practices might also explain national scale contrasts in apparent site densities. Arable farming is more common in the south and east of Britain, and 'rural' population densities are also higher (a factor discussed below), with livestock a more common land-use in the north and west of England as a whole. Smith and Openshaw' s

20

(1990) two zones of population, in the north and south (with the later including the area around the Humber estuary), might thus be a factor of land-use practices rather than Mesolithic occupation densities. Detailed analysis may be necessary however to determine if this is the case.

Clearly, although the composition and character of upland sites, if not their distribution, is often considered to be free from bias, the effect of different individuals, and of the unique character of upland excavations, can affect both the character and composition of upland assemblages.

Another land-use effect, which in this case affects the recovery (rather than the visibility) of sites, is the densities of people living near any area. Human populations provide a reservoir of potential collectors, who might find and record sites and relay this information to sites and monuments records or publications. Actually most 'flint collectors' are 'country folk', however the densities of populations in 'rural' areas are much higher near large cities, providing another potential explanation for the high densities of sites recovered in the Central and South Pennines. Population densities also affect the numbers of sites recovered through professional archaeological studies, as many of these are carried out in advance of development work, which tends to concentrate in big cities. On the other hand, the construction of these cities themselves has often obliterated any evidence for Mesolithic sites. Young (1986: 218-224) cites the influence of a lack sites around Sunderland as an example of this process.

Several authors note the typically 'small' size of recorded upland sites. Both Mellars (1976) and Myers (1986; 1987) for example have drawn upon the small spatial extent of upland 'sites' in their interpretations. However there are several reasons why upland sites in contrast to their lowland counterparts may have been recorded as relatively 'small'. One immediately noticeable characteristic of upland excavations is that they tend to be 'incomplete' (Stonehouse 1990: 62), in the sense that artefact distributions continue beyond the area excavated (demonstrated for old excavations re-excavated in the Pennines, Spikins 1994). As noted above, the real distribution of artefacts in upland sites, albeit being many phases of activity, can be large. The extent of a 'site' is therefore largely a measure of the available time, difficulty of excavation and determination of the collector. In the Pennines two general factors may be acting to limit excavated site size. First, the nature of the overlying sediment can affect site 'size', since sites tend to be excavated at the edge of the peat margin (where artefacts are exposed) here peat depths increase rapidly away from the marginal face (by as much as 10cm per metre) making excavation increasingly time consuming. Secondly, wet, windy and misty weather conditions in the uplands are a major deterrent to prolonged excavation.

The influence of nearby populations is very variable at a local scale. First, because of the effect of access routes, roads and footpaths. Sites on Rombalds moor, for example, are concentrated along the main brow of the moor following the footpath. Secondly, because of the 'individual' factor. Certain individual flint collectors (or even professional archaeologists) can have a marked effect on site distributions. Young (1986: 218-224) notes that in Weardale sites are particularly concentrated near the home of one important collector (E. J. W. Hild yard), and Young and O'Sullivan illustrate the effect of several collectors on the distribution of sites in the North East of England (Young and O'Sullivan 1993). In the Pennines the influence of several key collectors, have had a clear effect, with sites excavated by Francis Buckley in the 1920s being concentrated near his home town of Marsden, for example. The effects of individuals on site distributions are often more complex than they might at first appear. For one thing the numbers of sites in any area will be affected by what any individual chooses to term a 'site'. Most sites in the Pennines are actually collections of a very small number of artefacts, often even single artefacts picked up in an ad hoc manner on a 'Sunday afternoon stroll' (figure 2.10). One individual might consider that each collection 'event' has recovered a 'site', whilst another might collect from the same 'site' for many years (Yarwood and Marriott 1994). Standards of accuracy in the location of sites are also very variable (Yarwood and Marriott 1994). Aside from the different densities of 'sites' which different methods of collection (and of erosion) might produce it is also very difficult to relate socalled 'sites' to real patterns of past activity. Excavations and test-pitting and auger surveys on Marsden moor have revealed that artefact distributions can be more or less continuous over a large area (about 250m 2 in the case of March Hill Carr) and relate not to one but to many overlapping phases of occupation.

Upland weather conditions combined with the 'human factor' may also affect site characteristics in other ways. The potential for noting and recording features in perpetual drizzle is low. In fact, a weatherproof covering was found to be essential to ensure adequate finds recovery when excavating sites on Marsden moor. Whilst features are traditionally rare on Pennine upland sites, under covered excavations an average of one feature every 5m 2 excavated was recorded (including five hearths, a stakehole and an undiagnosed feature). It is hence no surprise to discover that recorded upland sites are smaller in size and contain fewer features than lowland sites. Essentially, these characteristics may be more to do with the nature of upland excavation than with any contrasts in the types of upland and lowland activities in the past. The 'human factor' can also affect the recorded composition of assemblages. Almost all the sites which can be typologically dated, and almost all of the sites with identified microliths in the Pennine dataset studied here, for example, are clustered around Marsden moor. This is largely because of the care and attention paid by local flint collectors and amateur archaeologists, from Francis Buckley to more recently Pat Stonehouse, and their ability to identify and record different artefact types and to pass this information on to the official SMR record. Museum collections from actual excavations appear to be the best evidence available for variations in assemblage composition (and form the basis of the analyses by Jacobi 1976, Mellars 1976 and Myers 1986) but these may also be very biased. Francis Buckley, whose collections form the

bulk of these analyses, commonly selected the 'best pieces' from any assemblage to send to different museums, before leaving the remainder as the main record of the sites in local museums. Since no regional analysis of recently excavated assemblages has been carried out, interpretations are thus heavily dependent on the biased records of museum assemblages. Clearly the relationship between human and natural biases is complex and interrelated. It is not possible to 'factor out' different biases both because they are interWest Yorkshire Mesolithic Project - excavations on Marsden Moor. dependant, and because such Figure 2.16 detailed information (such on how frequently footpaths MARSDEN MOOR - A DETAILED STUDY. were used) would be needed. However, it may be possible to get closer to accounting for biases at a local scale, where it is possible to collect detailed information. Discussions at this Marsden moor (figure 2.16, and distribution of sites, figures scale are particularly important, since the common 2. 7 and 2.8) has been a key area in discussions about the preferences of 'sites' recorded at this scale have been taken landscape 'preferences' of Mesolithic sites, and is often used to imply a long term continuity of upland hunting activities as an illustration of these preferences (such as Simmons and of Mesolithic populations. A detailed study of the biases 1996: 34). One element of the research carried out by out affecting 'sites' at this scale, on Marsden moor, has been West Yorkshire Mesolithic Project (Spikins 1994; 1995b; carried out through West Yorkshire Mesolithic Project 1996) has been a detailed study of vegetation and erosion (Spikins 1996), the results of which are discussed below. patterns in this area. This study was carried out in order to address the importance of biasing factors affecting the surface recovery of Mesolithic material, and to provide information for future management of the moor. Marsden moor has a long history in Mesolithic research. It has been subject to the attention of collectors and archaeologists since the 1880s. The 'sites' of March Hill, March Hill Carr, Dan Clough, Lominot and Dean Clough are all clustered together, at the narrowest part of the Pennine watershed. Law and Horsfall were the first to mention March Hill, having collected artefacts from this moor in 1879 (Law and Horsfall 1882). Later, the work of Francis Buckley in the area in the 1920s was of major importance to the wider discipline (Buckley 1924 and unpubl.). Buckley was instrumental in the discovery of two phases of the Mesolithic (the Early and Late Mesolithic) which he termed the 'Broad Blade' and 'Narrow Blade' industries respectively. The importance of his excavations at Marsden moor are cited by Petch (1924), Clark (1932), Radley and Mellars (1964) and Jacobi (1976). Radiocarbon dates taken from Buckley's excavations have also been influential in the dating of the British Mesolithic (Switsur and Jacobi 1975; 1979) with four of the eleven sites dated being from Buckley's sites in this area. The whole area was intensively 'scoured' by collectors in the 1960s (Turner 1964, cited above). March Hill, Dan Clough, Dean Clough and Lominot are also key sites in local publications such as Barnes (1982), Williams (1985), Stonehouse (1987; 1990), as well as national research

22

However, different types of erosion tend to occur in different conditions of altitude, slope, vegetation cover, and rainfall. The two most important types of peat erosion are dissection systems and marginal face erosion (Bower 1960).

syntheses and assemblage analyses such as Jacobi (1973; 1976) Mellars (1976) and Myers (1986; 1987). The most recent references include Smith (1990) and Simmons (1996). The area has also been researched as part of the West Yorkshire Mesolithic Project with four field seasons running from 1993 to 1996. Given the intensity of collection at these sites, it is a reasonable assumption that biases on the visibility of sites are the main recent factor influencing recorded site distributions, making this area a particularly interesting area to study. Here at least there may be potential for accounting for the main factors of erosion which have biased site distributions.

The main type of erosion on the peat plateau, both at Marsden moor and more widely in the Pennines, are dissection systems (often termed Type I erosion). These tend to be present at the highest elevations. Erosion breaks up the peat body into haggs and groughs (termed reticulate erosion), which can also include a branched network of gullies on sloping ground. Dissection systems are familiar for example to walkers over Bleaklow in the South Pennines. This type of erosion does not necessarily reach the peat base but may in cases be restricted by a denser layer of peat at the base of the peat profile.

Marsden moor is typical of an upland Pennine landscape. The main body of the peat plateau is covered with vegetation characteristic of upland peat areas - wavy hair grass and cotton-grass, with matt grass in drier areas. The hill slopes are dominated by matt grass and moor rush lines stream beds. The peat areas themselves are eroded across the plateau and at the edges of the peat face, and streams running eastwest have also cut steep sided valleys and in some areas steep scree slopes. Modern human influence is clearly felt as the moor is crossed by a main road, the A640 which runs from Huddersfield to Manchester, and is criss-crossed by a number of footpaths, including the Pennine way (none of which coincides with the distributions of recorded sites). It is also a popular spot for tourists, especially since it is near to a viewpoint to the north on Cupwith Hill and because of the pub overlooking March Hill from the north, Buckstone' s Inn.

At the edge of the peat plateau, where a gentle peat covered slope suddenly steepens marginal face erosion occurs (Johnson 1957). This is the type of erosion shown at the edge of the peat face on the north and south-facing slopes of Dean Clough as well as the north and south facing slopes of Dan Clough and March Hill. In deep peat on gently sloping ground (steeper than 3-5°), steep-sided unbranched gullies can extend back from this peat margin for a few hundred metres (termed Type II erosion). Alternatively arcuate scars (with near vertical back walls on slopes steeper than 10°) can also develop. These latter types of erosion are often termed 'sheep scars' and are likely to be the result of sheep action (Yalden 1981).

At first inspection, the pattern of erosion in this area appears to confirm that the locations of sites represent real 'preferences' on the part of Mesolithic populations. As noted by Spikins (1993; 1995c) most sites have been recovered from locations with quite limited erosion, that is, the southfacing slopes at about 420m OD elevation. Erosion is much more extensive both at higher elevations (on the Pennine watershed) and on north-facing slopes which suffer from severe frost action (as they are colder, receiving less warmth from the sun).

A detailed survey of vegetation and erosion patterns across the valley of Dean Clough, and further surveys across the main peat plateau, at first confirmed the patterns noted above, with the densest concentrations of sites being found where erosion was limited, with few sites found where there was extensive erosion. However the explanations for this patterning were complex, as discussed below.

KEY

KEY

severe erosion

matt grass

revegetating erosion

wavy hair grass and crowberry

cottongrass water

wavy hair grass and cotton grass

wetland/

river vegetation

water

Figure 2.17

Vegetation Survey, main vegetation types.

2.18

23

Vegetation Survey, types of erosion.

A Detailed Survey

Figure 2.19

Vegetation on Dean Clough was recorded according to 30 different categories of plant communities, in north-south vegetation transects spaced at 20m intervals. The area covered was approximately 260m along the valley and 300m across the gradient. Figures 2.17 and 2.18 show a summary of the results of this vegetation survey with vegetation types plotted in 'on the ground' distances. The stream marked is Readycon Dene stream, and shows a clear association with the wetland vegetation. The survey results also show the typical concentration of matt grass on the steeper slopes of the valley sides (figure 2.17), and wavy hair grass and cotton-grass (marked as 'short grass', or with crowberry as 'crowberry') on the main body of peat. Matt grass also extends onto and above the marginal peat face, a reflection of the extent of grazing (matt grass is resistant to grazing by sheep).

North-facing slope of Dean Clough.

Erosion in this area is concentrated in two locations, the higher parts of the peat plateau (studied in a different survey), illustrated in figure 2.20 and the marginal peat face within the valley, figure 2.19 and 2.21. The survey revealed a distinct contrast between the north and south-facing slopes that is typical of Pennine valley systems (which predominantly run west-east).

Figure 2.20

On the north-facing slope the greatest extent of 'severe erosion' is recorded. This is largely because this slope receives the least warmth from the sun and the surface is subject to intense frost action, especially over the winter months. In these areas the marginal peat face erosion is continuous with deep linear groughs along lines of streams and water movement from the peat plateau (Type II erosion), figure 2.19. The eroded peat extends to the plateau beyond (visible in figure 2.20) which is also heavily eroded (Type I erosion).

Central Plateau.

On the south-facing slope, there are more small patches of 'regenerating erosion' (i.e. where vegetative growth is apparent), figure 2.18. This is where marginal sheep scars have exposed a flat area at the mineral soil level (rather than extending further into the sandstone) and this matrix provides better nutrients for vegetation, as well as receiving more sunlight, than the exposed peat of the north-facing slopes. This re-vegetation can be seen clearly on figure 2.21 (showing the original exposed area which has also been subject to damage from collectors). Crowberry survives here with purple moor grass and cotton-grass on the drier edges of the sheep scars (see figure 2.17). Crowberry is replaced by wavy hair grass and cotton-grass, without crowberry, on the north-facing slope.

marginal face sheep scar Figure 2.21

South-facing slope of Dean Clough.

It is possible to determine the level of Mesolithic artefacts within the peat profile quite accurately. GIS-based models of the three-dimensional artefact location according to the stratigraphy demonstrate a clear clustering of artefact locations at the mineral soil - peat interface (Spikins 1994; Spikins, Ayestaran and Conneller 1995). The effect of different erosion types, which occur in different topographic situations, on exposing this level are summarised below, with reference to figure 2.22.

24

SOUTH FACING SLOPE

UPLAND PLATEAU

NORTH FACING SLOPE

'sheep scars'

Type 1 dissection gullying

Type 2 dissection gullying

wind erosion peat

paleosol sandstone Figure 2.22

Model of Erosion types.

The Influence of Erosion Type on Finds Recovery

disturb the su,face and frost, wind and rain act upon it. Stones are kicked away downslope from the apron onto the grass so covering and killing it. In places large areas of thin soil are being stripped from the underlying weathered rock Evans (1992: 54)

On north-facing slopes erosion is severe due to frost action, and with subsequent gullying (Type I erosion), mineral soil can be exposed. Artefacts may be recovered in a narrow broken band at the marginal peat face, although the extent of surrounding eroded peat and its fluidity means that exposed mineral soil is often re-covered by a re-deposited peat layer (see figure 2.19).

The effects of sheepscars in eroding peat and exposing finds might at first seem minor. Sheepscars are not the most visible types of erosion, particularly because this type of erosion exposes mineral soil, which re-vegetates quickly, rather than exposing large areas of damp dark peat. However, the total area of sheepscars in any location can be large. Evans (1977) attempted to classify the specific causes of soil erosion in a specific valley, Hey Clough, South Yorkshire, concluding that sheepscars created 35% of the total erosion, with some single scars reaching sizes of 20-30m 2 in area.

On the central plateau peat can reach depths of up to 8m. Thus even when extensive, plateau peat erosion does not necessarily extend to the finds level. Severe erosion (Type II), especially in areas which have suffered the effects of moorland fires, can reach the artefact level and quite frequently extends horizontally exposing large areas of sandstone matrix with steep sided gullies (see figure 2.20). Whilst finds are exposed in these situations it is difficult to recover artefacts since the small size of most artefacts, especially diagnostic elements such as microliths (often only about 1cm long) makes them difficult to separate from eroded sandstone.

A closer consideration of the effect of erosion in exposing artefact levels reveals several important patterns. We would expect the most severe erosion to be at valley heads at pollutants concentrate in these areas (Lee 1981). However, it is also clear from the above discussion that the visibility of artefacts is not greatest at highest elevations where peat erosion is most severe, as had been assumed, but is instead greatest at lower elevations (about 420m OD) at the marginal peat face. Furthermore, artefacts are also most visible on south rather than north-facing slopes.

On south-facing slopes erosion is less severe, however the main agent of erosion in these locations is sheep. Sheep prefer to graze and ruminate on warmer south-facing slopes. They encourage erosion, not only by trampling and by consuming vegetation, but also by sheltering in breaks in vegetation at the marginal peat face. The action of sheep (in creating so called 'sheep scars') is particularly important since it exposes a large area of mineral soil at the finds level (see figure 2.21), and since sheep instinctively prefer to have a clear view of their surroundings, these locations also tend to have a 'good view'.

Effectively, the optimal recovery locations for sites, according to erosion patterns should be at pleateau edges, on south-facing slopes, especially at valley heads, and in locations with a wide area of view. These are exactly the types of location interpreted as relating to Mesolithic 'preferences' by Radley and Marshall (1963: 96), Jacobi (1976; 1978: 32), Barnes (1982: 25), Spikins (1993; 1995c), Simmons (1996: 33-4) and Kvamme and Jochim (1985). Moreover, further confirmation that the locations of sites are largely a reflection of the location of the marginal peat face have been revealed at Marsden moor, where recorded 'sites' have been related to past marginal peat face erosion patterns (through contour surveys and surveys of vegetation

Evans describes the action of sheep in creasing 'sheepscars' in detail. Sheepscars are formed which comprise a backwall and scar apron (Evans 1977). The backwall continually retreats because sheep rub themselves on it or it is broken down by their hooves, whilst the apron extends because the sheep 25

Figure 2.23

Recorded marginal peat face erosion in the southern Pennines and the distribution of recorded sites.

characteristic of previously eroded areas) even where presently the locations of these sites are not eroding. (Spikins 1995b; 1996).

below the average i.e. 453m ±48m). This also appears to correspond to the limits of the site distribution. That the 'lack' of sites at the highest elevations could be a result of differential erosion patterns should come as no surprise. It has already been suggested that erosion may be creating false patterns in site distributions, and that sites may be hidden in certain upland areas (Raistrick 1932; Garton 1987). Moreover, from an ecological perspective, there is no particular reason to assume that Mesolithic populations were constrained to low uplands. In the Alps, apparent long-term 'domestic' occupation sites, where children's teeth have been recovered, have been found at elevations as high as 1,9002,300m in the PreBoreal (Broglio 1996: 42). Given the biases from differential erosion operating at high elevations, and those from different land-use practices in the mid-uplands, there is little evidence for a distinction between upland and lowland sites on the basis of site distributions.

The influence of differential erosion on site distributions at the local scale, has far-reaching implications. At the local scale itself, there may be more sites located on south-facing slopes at around 420m OD, where there is a good view, in fact this preference would not be an unreasonable one to expect from hunter-gatherers. However, there is no evidence that these were the preferred spots, nor, more significantly, that the same spots were preferred for millennia. At the regional scale, the biases towards a specific elevation also suggest that, not only is there a marked distinction between upland and lowland sites which is quite probably a result of differential erosion, but moreover that the lack of upland sites above this elevation may also be a result of different erosion patterns. The location of the marginal peat face in the South Pennines (Phillips, Y alden and Tallis 1981), figure 2.23 , for example, suggests that this type of erosion is a more important influence on distribution than severe erosion (figure 2.6). Using the elevation of the known marginal peat face in the Southern Pennines a rough idea of the location of this marginal face in the Central Pennines, where the main densities of sites have been found, can be mapped, figure 2.24 (using here one standard deviation of the variation in elevation of the marginal peat face above and

There are also implications for broader ideas about adaptations. Given the patterns outlined above, it appears that a narrow 'window' of increased artefact visibility and preferential recovery could exist in many upland regions (where sheep are a major agent of erosion of peat covered areas). The distribution of sites on Marsden moor, and in other areas, is thus a 'false pattern' if we are looking to interpret human activity. Considering the effects of biases thus casts doubt on interpretations of upland hunting, and of

26

.

.

..•

:·::-i'Ll ..---■-··~ •.:=

. . .

..•

.

.

.

.

.

,

.

0

Figure 2.24

5

10 km

Model of marginal face erosion in the Pennines and the distribution of recorded sites.

a continuity of landscape use between the Early and Late Mesolithic or between the Late Mesolithic and Neolithic which depend on the basis of sites clustering or common 'preferences'. From the large scale, to the medium scale, to the small scale, the effect of biases has been found to be much more pervasive than was at first assumed. There initially appeared to be several clear patterns in the temporal and spatial distribution of sites in northern England, which might provide a means of 'building up' a model of Mesolithic adaptations, on the basis of the archaeological evidence for population and settlement. Patterns included clear increases in the numbers of sites, and by implication in population numbers, through time, which tied in with changes in the apparent settlement systems of different hunter-gatherer

groups. There also appeared to be evidence for broad distinctions between different social groups at the national scale, for specific settlement systems at a regional scale, as well as for distinct upland and lowland patterns of activity, and at the local scale for a continuity of use of the uplands. Many of these interpretations have been seen to be built on 'shaky ground'. Interpretations relying dominantly on the distributions of sites are seriously affected by biases introduced from variations in the natural physical landscape, as well as from the 'human factor'. However, even patterns in the composition or character of different sites are affected by biases, introduced by the character of an individual collector or excavator as well as by the characteristics of upland environments. Other types of bias, which also affect interpretations, such as common preconceptions about the

structure of settlement systems, are considered in a later chapter dealing with methods of interpretation (chapter four). It is already clear here however that the effects of biases make a 'top-down' approach to Mesolithic adaptations severely limited.

CONCLUSIONS There are two main particularly important points which we can draw from the above discussion. First, that the influence of biases is pervasive. The effects of biases on the distribution of sites are much more far-reaching than we might expect and mean that interpretations drawing on site distributions have little firm footing. Even interpretations based on the composition of assemblages, rather than simply on the distribution of sites, are affected by many different biases. Secondly, the mechanisms of bias are extremely complex. It required detailed analysis to 'untangle' the biases that operate in a local landscape, biases which have a serious effect on possible evidence for a continuity of upland activities and occupation. It may not be possible to simply identify biases and account for them. Essentially, the main conclusion is that the archaeological evidence for large scale patterns in northern England alone, as it stands, is insufficient to 'build up' a model of Mesolithic adaptations. There may be some patterning which does relate to past adaptations, such as the stylistic zones of different microlith forms, or the distinctive division between Early and Late Mesolithic assemblages, however these patterns are far from easy to interpret, without a better understanding of the context of Mesolithic environments, and a framework within which to place interpretations. Moreover, given the possibility of unexplored biases at work they should be used with care. In the next chapter, chapter three, the basis for the alternative 'bottom up' approach to Mesolithic adaptations, the available evidence for subsistence resources, is considered in detail.

28

CHAPTER THREE

Working 'Up' from Resources

ABSTRACT Interpretations of subsistence practices are a crucial element of models moving from environments, to resources, to population and settlement. However, the history of research into subsistence practices has tended to be one of isolating 'the resource' which is the most important or staple food, and which may define settlement patterns. In many cases, the proposed resource or suite of resources has been selected based on common assumptions about Mesolithic lifestyles, rather than any secure evidence. A detailed discussion of the range of possible subsistence resources demonstrates that focusing on any single resource in this way can be somewhat problematic. For one thing, any one of a number of resources could in particular circumstances have been a staple in Mesolithic Britain. For another, it may not be the 'staple' resource which is the most important archaeologically, population numbers are more normally defined by the resources available at the 'poor season' for example and certain resources, only available at specific locations or seasons, such as seasonally abundant salmon 'runs', may exert a major 'pull' on settlement. Developing a better understanding of possible Mesolithic subsistence strategies will depend on considering a number of characteristics, rather than just abundance, which are clearly important in determining how resources are exploited. Key important characteristics can be defined. However, due to the lack of understanding of factors such as past environments, the role of methods of resource exploitation in determining how resources were exploited, and the importance of the past history of societies in defining subsistence and settlement patterns, any absolute model of subsistence is probably beyond reach. Explicit methods for determining subsistence practices are discussed in chapter four, but it is clear that only a very 'coarse-grain' approach, concentrating on contrasts between different types of resource environment, would minimise the effects of a lack of evidence for specific subsistence resources and the pervasive influence of popular pre-conceptions of 'key' resources.

INTRODUCTION conditions discussed in chapter two. Evidence for the exploitation of resources in neighbouring regions may be relevant, although clearly this needs to be used with care.

Probable subsistence practices tend to be an integral part of the 'first chapter' of any study of the British Mesolithic. Subsistence often becomes the bottom rung of a ladder of inference used to build up an understanding of past huntergatherer societies, and the interpretations of changes in settlement and technology are based on ideas of changing subsistence. This is, at least in part, due to ample support for the importance of subsistence patterns in defining recorded hunter-gatherer demography and settlement (Birdsell 1953; Baumhoff 1963; Casteel 1972; Thomas 1981; Kelly 1995). It is particularly noticeable, for example, that abundant and reliable resources, most particularly coastal resources, are intimately linked with social complexity (Perlman 1980, Rowley-Conwy 1983; Keeley 1988) with accompanying characteristics such as storage, sedentism and extensive settlements (Price and Brown 1985: 11; Soffer 1989; Testart 1982; Yesner 1983). It is often suggested that 'complex hunters are likely to develop where a suitable array of migratory resources are available' Rowley-Conwy (1983: 118).

Other lines of sources of evidence may also be useful. One key element to consider is the availability of different food resources - how abundant they may have been and where and when they would have been available. The main source of evidence for this element is from modem environments similar to those of the Mesolithic. Records of the exploitation of different resources by modem hunter-gatherers, or historical records of traditional uses in the past may also provide a better understanding of how resources might have been exploited or the roles they might have played in subsistence. Interpretations of subsistence practices in the Mesolithic have changed as perspectives on the period have altered, and as new evidence has become available. Early interpretations of subsistence practices tended to concentrate on the 'most obvious' or most visible resources. Early this century, for example, the presence of Mesolithic shell middens prompted assumptions that Mesolithic populations eeked out a meagre existence (Clark 1932), living largely on shellfish (as discussed in chapter one). The recovery of evidence for a range of other resources, particularly large game animals, prompted more carefully considered interpretations of Mesolithic subsistence practices (such as Clark 1972). However, though more considered, these interpretations were also biased by the nature of the evidence. Later interpretations suggested that less 'visible' resources which are rarely preserved, such as plant foods, were more likely to be the key elements in subsistence strategies. The most recent development (in the 1980s) has been the attention drawn to marine and coastal resources as determinants of social complexity. Though applying a greater appreciation of the biases in the archaeological record, almost all interpretations have tended to be geared towards identifying a single key resource, or group of resources, which would be of major importance. A central problem is that, although the selection of these specific resources appears to be lead by archaeological evidence or environmental or ecological models, it is actually very dependant on the 'fashionable' ideas of the time.

As well as being important in a direct sense, as one of the key factors influencing population and settlement, interpretations about subsistence practices also influence our perceptions of the Mesolithic. Clark (1932, cited above) for example clearly saw a diet of shellfish, roots and berries as indicative of a 'poverty-stricken' state. Subsistence strategies based on hunting of large game animals - such as deer or aurochs - are on the other hand seen as more noble or affluent. In fact, archaeologists themselves often prefer to research and make interpretations about the 'dynamic' hunting of large land mammals rather than shellfish or plant collection, and this is one of the factors which has tended to restrict or bias the range of interpretations made about the past subsistence (Ehrenberg 1989; Watson and Kennedy 1991). There are several key elements to even a basic understanding of past subsistence strategies. The first question to address is simply which resources were exploited. Beyond this, the role of any resource in relation to other resources, its means of exploitation, and how the use of different resources may have changed seasonally, are also important questions to consider. Even these basic queries are difficult to address given the paucity of direct evidence for subsistence, and the problematic nature of evidence from analogous environments or societies.

The discussion below is not intended as a complete review of Mesolithic subsistence resources, but rather as an illustration of the range of resources available to Mesolithic populations, and to demonstrate that the question of key subsistence resources, and of subsistence practices, is complex, and despite a series of proposals, is as yet unresolved. To make this discussion more approachable, resources are grouped into 'common' categories, rather than by taxonomy, and common names are used for all resources, with the Latin names referenced in Appendix A. The locations of the archaeological sites mentioned are illustrated in figure 3.1.

The most obvious source of evidence for past subsistence is direct evidence of exploitation from archaeological sites. Examples of this kind of direct evidence might include faunal (bone) remains from the hunting and processing of game animals, charred nutshells from roasting nuts or the remains of discarded shellfish shells. However, direct evidence for the exploitation of specific resources is rare, especially in northern England with the poor preservation

30

Sites in northern England StC B ThC Cs

Star Carr and Barry's Island Blubberhouses Moor Thorpe Common Cass ny Hawin

Sites in the British Isles I Mo F 0 Ca R Mc CuW Th MtS N LB

Islay Morton Friarton Oronsay Carding Mill Bay Risga McArthur' s Cave Culverwell Thatcham Mount Sandel Newferry LoughBoora

Sites in Western EuroRe Rk Sk Er 0g As S!il T 0Ly Ha Mu Ar

Ringkloster Skateholm Erteb!illle 0gaarde Aggersund S!illager Teviec 0lbyLyng Havn!il Mullerup Agerod Tybrind Vig

Figure 3.1

Sites in northern England mentioned in chapter three.

SUBSISTENCE RESOURCES

THE SPREAD OF FLORAAND FAUNA INTOTHE BRITISHISLES

land mammals are such a visible element of past subsistence practices, and moreover a resource that is still 'hunted' today, many interpretations have also been based on a dominant role for these resources in subsistence. The influence of large game in general, and red deer in particular, on subsistence and settlement, has formed the basis for several key discussions of Mesolithic settlement. The first explicit model was put forward by Clark (1972), as cited above. On the basis of the most frequent faunal remains at Star Carr, Clark suggested that populations in Britain in the Early Mesolithic would have been largely dependant on red deer. He also used red deer ecology to suggest patterns of Mesolithic settlement. Clark's ideas were a major influence on the idea of upland 'hunting sites' contrasting with lowland 'base camps' noted in chapter two. Thus, Jacobi (1978) and Myers (1986; 1989) also assumed a dominant role for red deer in subsistence and made similar interpretations about settlement patterns (which are discussed in more detail in chapter four).

The Early and mid-Holocene marked a period of substantial changes in plant and animal communities within the British Isles. Several lines of evidence (addressed in detail in chapter five) point to a rapid rise in temperature in northern Europe after the end of the last glaciation. Plant communities will however have taken time to spread from glacial refugia, with animal species thus following these changing habitats, rather than temperature changes per se. Nonetheless by the start of the Early Mesolithic (taken here as 10,000bp, approximately also the time when 'Mesolithic' toolkits begin to appear) 'open' woodland communities of plants and animals would have replaced grassland over much of northern England. With time, as slower moving tree species arrived in the British Isles, woodland communities will have become dense - more 'closed' -, and plants and animals better adapted to these closed woodland conditions would have gradually displaced more 'open' woodland species. The nature of plant and animal communities has also been affected by the severance of the 'landbridge', between Britain and the continent (and between Britain and Ireland) and the influence of humans themselves, through exploitation, and through accidental or deliberate introduction or encouragement of particular species.

Deer Since Clark's (1972) model red deer in particular have played a key role in interpretations of Mesolithic subsistence strategies. One of the most famous statements made about Mesolithic subsistence is that one red deer carcass is equivalent in energetic terms to 52,267 oysters (Bailey 1978: 39), the resource previously assumed to typify Mesolithic economies. Bringing 'home' a large kill, such as a deer, is clearly important not only in practical terms, but also confers status on the hunter, who shares the meat with the rest of the group and gains prestige (Kelly 1995; Mithen 1990).

Unfortunately our knowledge of the date of arrival of different species of plants and animals and their relative availability in Mesolithic environments is poor. The best evidence concerns the spread of tree species (recorded in pollen cores), although evidence also exists from pollen diagrams for the spread of understorey and herb flora (largely much faster than slow moving tree species). Evidence for the spread and availability of large fauna is however much more sparse, and largely dependant on assemblages from archaeological sites (which present a biased selection of available species).

Two main types of deer, red deer and roe deer, would have been available to Mesolithic hunters. Red deer would also have been an important resource not only for meat but also for hides and for antlers, whereas roe deer antlers are usually too small to have been valuable. Both roe deer as well as red deer bones are typically recovered at Mesolithic sites where faunal remains are preserved. In northern England, red and roe deer remains were recovered from Star Carr (Fraser and King 1954), and in the rest of the British Isles, in Scotland at Morton, Fife (Coles 1971), Oronsay (Grigson and Mellars 1987), Carding Mill Bay (Hamilton-Dyer and McCormick 1993), and in southern England at Thatcham (Wymer and Churchill 1962; Wymer 1991: 27). In fact, in 1972, Jarman noted that 98% of European Mesolithic sites with faunal remains had been found to contain red deer remains (Jarman 1972). Although some analyses have shifted attention away from the red deer component of faunal remains at Star Carr itself (Legge and Rowley-Conwy 1988; 1989), recent analysis of a site very near to Star Carr, Barry's Island (Rowley-Conwy 1994), with a high proportion of red deer bones, has maintained the perceived importance of deer to Mesolithic populations.

Models of the spread of tree species as a basis for understanding changing resources, are discussed in detail in chapters five and six. Until then however an broad idea of the availability of different species of plants and animals is best given through a general discussion of what evidence exists for availability and exploitation of different types of resources. Evidence for changes in these resources through time is included where available.

LARGELANDMAMMALS Alhough the date of arrival (and in some cases the extinction) of large mammals is often difficult to define , the large land mammal component of past food resources is the most 'visible' in the archaeological record, and large mammals present the best evidence for use of resources. This is because faunal remains are the most common surviving direct element of subsistence practices, and of these remains the large mammal elements are most likely to survive postdepositional disturbance in an identifiable state. Since large

The way in which deer would have been hunted is largely determined by the ecology and behaviour of deer populations. In forested environments, the size of deer groups varies with season and vegetation, with the sexes typically remaining separate for most of the year but coming together for the autumn rut (Jochim 1976: 105). Red deer and roe deer have slightly different behaviour patterns. Red deer 3.2

aggregate into groups of several does and young (and thus several animals might be killed at once). Roe deer on the other hand, live in smaller, primarily family groups (Jochim 1976: 106), in contrast to red deer they prefer to feed on the shrub layer in forests rather than on grasses. The difficulties encountered in hunting red deer and the success of any hunt are essentially determined by the density of deer populations. The densities of both deer species depend on the density of available forage as well as on the intensity of predation. Roe deer are likely to have existed at lower densities relative to red deer and wild boar in Mesolithic forests as they suffer more from predation, and also compete poorly with red deer and wild boar (Jochim 1976: 102-3). Since deer thrive in open woodland, forests would have gradually become a less suitable habitat for deer throughout the Mesolithic as forest density and shade increased (Jochim 1976: 101) and as forage reduced in density. As noted by Keene 'contrary to popular belief, the climax forest is not good deer habitat' (Keene 1981: 101). Nonetheless, although deer has a prime role in interpretations of Mesolithic subsistence, especially in northern England, recently other large game animals have been recognised as being potentially as important. Legge and Rowley-Conwy (1988; 1989), for example, remarked on the potential importance of other large game at Star Carr. In other areas of Europe, analysis of faunal remains also revealed the importance of other species, particularly wild boar. In Denmark, Rowley-Conwy's (1984) analysis of Ertebjljlle faunas suggested that wild boar were particularly important in Late Mesolithic subsistence practices, with recorded kills of about 50% wild boar, 30% red deer and 20% roe deer. Other deer which may have formed an element of Mesolithic economies include reindeer and elk. Although a famous, and vital resource for Upper Palaeolithic hunters, reindeer, if available in the Early phases of the Early Mesolithic would have been in decline. Elk however would certainly have been available in the Early Mesolithic, with elk populations reaching their maximum in the early stages of forest succession and declining rapidly with the spread of closed woodland (apparently becoming extinct a the end of the Boreal (Jones and Keen 1993). Elk is the largest of the cervids, with an adult male weighing up to 500kg (Chapman 1975). Remains of elk have been found in northern England at Star Carr (Legge and Rowley-Conwy 1988, 1989). They feed on a broad spectrum of shrubs and herbs and also aquatic plants (Chapman 1975: 41). Although much less numerous than red or roe deer even in open forests, elk may have been easy to hunt, being very timid and easy to approach (Jochim 1976: 98).

Wild Boar Though smaller than deer, wild boar may have been a major resource, especially as they should have been more common in postglacial forests (Jonsson 1995: 152), particularly dense oak forests, and additionally wild boar populations can sustain much higher kill rates than can deer. Whilst wild boar may have been important for meat though, the skin is unsuitable for use for leather because of the high fat content and bristle penetration through the skin (Wijngaarden-Bakker 1989).

Wild boar is second only to deer as a component of most British faunal assemblages. Remains of wild boar have been recovered in northern England at Star Carr (Legge and Rowley-Conwy 1988; 1989), and in the rest of the British Isles at Oronsay (Grigson and Mellars 1987), Morton (Coles 1971) and Carding Mill Bay (Hamilton-Dyer and McCormick 1993) in Scotland and at Thatcham (Wymer 1991) in southern England. At Lough Boora in central Ireland, and Mount Sandel in northern Ireland, 98% of the mammal remains come from wild boar (Wijngaarden-Bakker 1989: 127), although the dominance of wild boar is partly a reflection of the lack of deer in Ireland. Wild boar is also a consistent component of most faunal assemblages in the Scandinavian Mesolithic. A further important point is that, whilst deer densities would clearly have declined during the Mesolithic, as forest density and shading increased, wild boar would have profited from a spread of oak forest since acorns (alongside roots and grasses) are an important staple resource for wild boar populations, especially prior to the winter months. Tilley (1979: 24) refers to historical accounts (in Howes 1948: 173) of acorns being a primary autumnal food on which wild boar were fattened in central Portugal. Male wild boar are solitary, but during the rut in November and December they join the female groups, which with several females and young can then range from groups of 6 to 50 (Jochim 1976: 106). These large groups may have provided an important resource. The age/sex patterns of wild boar hunted (adult female animals and juveniles) suggests that Mesolithic hunters at Lough Boora and Mount Sandel took advantage of the reduced mobility of sows with young (WijngaardenBakker 1989). At Scandinavian sites it has been suggested that wild boar may have been semi-domesticated, perhaps not permanently penned but foraging around human settlements (RowleyConwy pers. comm.), although no suggestions of this nature have been made for the British material. Defining domestication and identifying this in the faunal record is a complex issue. Other large game, particularly elk and aurochs, may have been less abundant, but being much larger than deer or wild boar may still have been important in Mesolithic subsistence practices.

Aurochs Remains of aurochs (wild cattle) have been found in northern England at Star Carr (Legge and Rowley-Conwy 1988, 1989), Barry's Island (Rowley-Conwy 1994) and in the rest of the British Isles at Thatcham (Wymer 1991) and Morton (Coles 1971). Like elk, aurochs are also interpreted as largely an animal of grasslands and open forests, being discouraged by denser Atlantic forests with a lack of grazing (Jochim 1976: 97, after Waterbolk 1968), although they continued to be present in these latter environments. Unfortunately, given that no natural aurochs populations survive it is extremely difficult to make interpretations about their ecology or habits (Legge and Rowley-Conwy 1988: 19).

of large land mammals is complicated by the fact that hunters are often opportunists, bringing home whatever type of game - large or small - that is encountered.

Horse Horse, again an animal of more open grasslands, was 'rare in the postglacial of Britain but not unknown' (Rowley-Conwy 1994: 3). Horse bones have been recovered from Barry's Island in northern England (Rowley-Conwy 1994) and Thatcham in southern England (Wymer 1991).

Hence as can be seen, the role of large land mammals in Mesolithic subsistence may have been inflated, not only by the relatively high visibility of large land mammal fauna on archaeological sites, but also by preconceptions about past subsistence practices as well as by the nature of the ecological and ethnographic evidence for resource exploitation.

Brown bear A vertebra of brown bear was recovered from the north-east of northern England at Star Carr (Noe-Nygaard 1983), but Jochim (1976: 99) considers that in general terms in Mesolithic forests bear would have been too rare to support a significant pattern of exploitation. Noe-Nygaard remarks on the decline of the species from Boreal to Atlantic times partly due to vegetation changes and also the isolation of Britain from the continent preventing new immigration. The occasional exploitation of brown bear in autumn and winter, when they have large fat reserves, may nonetheless have provided a welcome source of fat, as well as and meat and skins (Charles 1997). Nonetheless, bear has never been considered a major resource in any period.

The problems with ethnographic and ecological evidence for the use of large land mammals are subtle but important. Analogies between modem coniferous forests and Mesolithic woodlands (especially those in the Early Mesolithic) form the basis for estimates of large game availability. Likewise, subsistence patterns of hunter-gatherer groups living in coniferous forests (especially the boreal forests of the Canadian sub-Arctic) form the basis for Mesolithic subsistence and settlement practices (such as Price 1973, discussed in chapter four). However, today, expanses of coniferous woodlands are largely present in the cold climates of high latitude regions (such as the Canadian Arctic) and at high altitudes. This means that these environments are a poor analogy for those in the Early Mesolithic, since Early Mesolithic climates would have been similar to those at present (Mayewski et al. 1996) , with a restricted range of tree species present only because of slow rates of expansion of forest trees from glacial refugia (discussed in chapter five). Large game animals are the major resource for boreal forest groups in cold climate coniferous forests, but in Mesolithic forests, with abundant other resources available, this is less clearly the case.

DISCUSSION

Clearly different large land mammals species have very different characteristics and the relative importance of different species is a complex issue. In human terms it is not just the density of different species, or characteristics such as meat weights and other 'returns' such as hides and antler, which affect decisions about which species to target, but also how difficult species are to hunt and how frequently they are encountered. The paucity of archaeological sites with preserved faunal remains is legendary (especially in northern England where the two main sites, Star Carr, and Barry's Island, give conflicting evidence on seasonality and the relative importance of different species). The relative importance of different large land mammal species, or large land mammals in general, is thus largely a matter more of speculation than of informed conclusions.

Whatever their relative importance in diets however, large mammal resources are also important in that the study of faunal remains can potentially tell us about the season of occupation of archaeological sites, and thus potentially the season of exploitation of different large mammal resources. The season of occupation of many of the Mesolithic sites in Scandinavia which have yielded large faunal assemblages can often be determined with a reasonable degree of accuracy. In northern England however there are only two notable sites with large faunal assemblages, Star Carr and Barry's Island. Neither of these can give us a reliable assessment of the season of occupation of the area, the Vale of Pickering. For one thing there has been considerable disagreement about the season and function of occupation at Star Carr, (Legge and Rowley-Conwy 1988). Star Carr has been seen alternatively as a winter 'base camp' (Clark 1972), a butchery station or kill site (Caulfield 1978) a lakeside antler and skin working site (Pitts 1979) and a refuse dump from a nearby base camp (Price 1982), with occupation argued to be in late spring and summer on the basis of faunal remains (Legge and Rowley-Conwy 1988) or spring on the basis of charred plant buds (Day 1993). The faunal remains from Barry's Island (Rowley-Conwy 1994) suggests a winter occupation, even though the site is very near to Star Carr, making any broad determination of a 'general' season of occupation of this area, the only region with detailed faunal assemblages, almost impossible. The direct relevance of the archaeological evidence of large land mammals to

Whilst opinions vary on the importance and exploitation patterns of different species, the dominant role of large land mammals in general is often seen as a logical conclusion given their 'attractiveness' as a resource. Killing a large mammal such as a deer supplies far more meat than killing a smaller animal (estimates put the meat of one red deer as feeding five people for ten days without needing to consume any other foodstuff). Moreover, meat from large game animals is also often a preferred food, Lee (1968: 41) for example records that the !Kung 'eat as much vegetable food as they need and as much meat as they can'. Despite the attention that is often focused on meat from large animals however, both by archaeologists and by hunter-gatherers themselves, the actual contribution to diet of this resource for many known hunter-gatherers is often low (Kelly 1995). Large animals are difficult to catch and also, being unpredictable in their habits and precise location, are an unreliable resource (compared to fish or nuts for example). Studies of ethnographically documented hunters show that when specifically hunting large land mammals they often bring home smaller species or return empty-handed (Mithen 1987; 1989; 1990; Kelly 1995). In fact, the issue of the role

34

discussions of settlement patterns in this region is thus very limited.

SMALL MAMMALS

Small game resources are rarely credited with any significant role in Mesolithic subsistence practices. However, whilst being in smaller 'packages' than large game, small game can, in certain circumstances, provide a more important contribution to diets. This is partly because they are simply more abundant than large game (and therefore encountered more often when hunting) and also because they can be easily caught in traps and snares which require minimum energy investment. In forested environments, such as regions in Tasmania, small rather than large game are the primary inland source of meat (Lourandos 1997) Small game in open woodland environments would include small herbivores such as hares, squirrels and hedgehog, as well as predators, such as badger, wolf, fox, wild cat and pine marten (which are often seen primarily as a source of hides but can also be a source of meat). Small mammals such as hares, squirrels and hedgehogs, as well as possibly being important in 'poor' seasons, might also play an important role after periods of extreme climatic conditions (particularly cold winters or dry summers for example) since they reproduce rapidly and recover quickly from population decline, rising back to high densities soon after a poor year or series of poor years (Flowerdew 1987). The density of small herbivores and omnivores is generally dependant on available forage and small insect life (or, in the case of squirrels, nut production). However the relationship between the populations of predators and their prey can be complex and subject to multi-annual fluctuations (Flowerdew 1987; Mithen 1990). The main hare species found in Mesolithic Britain is likely to have been brown hare (rather than arctic hare), (Mayhew 1975). Hare bones were recovered from Mount Sandel in Ireland (Woodman 1978; 1985b). Hares prefer open environments, but are also found in forested situations, particularly boreal forests (Charles 1997). In contrast, squirrel densities, being largely dependant on nut production, would have been highest in open oak forest. Keene (1976) notes the hunter-gatherer populations in the Great Lakes of North America frequently exploited squirrels by catching them in traps or snares. Hedgehogs are perhaps one of the most unusual resources which might have been exploited in the Mesolithic. Hedgehogs are distinctive however since they have quite fatty meat, which would be attractive at times when the meat of other game resources was lean. There is also some possible evidence of their exploitation. The midden at Morton, for example, contained the remains of a hedgehog (Coles 1971; Smith 1990: 145). Though hedgehogs were unlikely to be found at high densities, Jonsson (1995: 152) suggests that there were a valued resource. He suggests that the spread of the hedgehog (which requires open landscape and is hindered by hilly terrain) in Sweden was deliberately encouraged by human populations. Hedgehog certainly appears to have been introduced by humans to the island of Gotland in the Baltic (Jonsson 1995: 152). Other very small mammals, such as mice, frogs etc. might have provided an additional resource at certain periods of the year, though given their small package size and difficulties in

35

capture they are unlikely to have been a significant subsistence resource.

predictable locations, although Renouf (1989) states that they are also easy to catch in spring when they are lethargic. At Star Carr, the beaver recovered were largely immature and it is possible that a whole family may have been killed whilst in their lodge.

With any small mammals it is always difficult to determine if the faunal remains recovered on archaeological on sites were the result of human exploitation or were animals which died when hibernating, burrowing, or were brought to the site by predators. For example, though remains of red squirrel were found at Carding Mill Bay in Scotland (McCormick and Buckland 1997) it is not clear if these bones were an exploited resource or an intrusive element. Conversely, small mammal food resources could have been important in subsistence, although they are often overlooked on archaeological sites precisely because they are not commonly considered to be an element of diets.

There are a number of historical and ethnographic examples of beaver exploitation. In northern Finland and Sweden, beaver were historically exploited in winter using cages, traps or nets (Broadbent 1979: 183). Beaver exploitation was also important amongst hunter-gatherer groups in northern North America, especially earlier this century, and particularly for pelts for the fur trade. Keene (1976: 99) after Kinietz (1965: 328) notes that among the Ottawa a good hunter would bring home as many as a dozen beavers in a day. Interestingly, many groups, such as the Ottawa, have been noted to deliberately conserve beaver stocks, by varying the lodges exploited each season and by leaving breeding pairs in a lodge (Keene 1976: 100 after Kinietz 1965: 237). However, Leacock (1954: 3) suggests that for the Montagnais of Canada, the conservation of beaver lodges might be a relatively recent practice adopted in historic times to preserve yields for the fur trade. However they were exploited, beaver lodges would probably be well known to Mesolithic populations and beaver could be especially valued as a source of fat in the lean months.

In contrast to the problems of identifying exploitation of small omnivores such as squirrels, we can be much more confident about the direct exploitation of predator species carnivores - from a number of British Mesolithic sites, since many of these bones have cut marks indicating butchery (Charles 1997). At Thatcham remains were recovered of badger, fox, wild cat, wolf and pine marten (Wymer 1991). Bones of pine marten, badger and red fox have also been recovered at Star Carr (Clark 1954; Smith 1990: 113). Nonetheless, although clearly exploited, the presence of carnivores in Mesolithic faunal assemblages does not necessarily signify their importance as a food resource. In Denmark, several specialist sites for the exploitation of furbearing mammals predominantly for their pelts have been discovered, such the site of Ringkloster (Rowley-Conwy 1984) which appears to have been a specialist camp for the exploitation of pine martens. Nonetheless, even if carnivorous mammals were not a preferred resource, they might still have been important in the 'poor season' or poor years.

Beavers became extinct in the twlfth century or later (Jones and Keen 1993). However, otters, sometimes confused with beavers (although otters and carnivores rather than rodents) survive to the present. Otters largely exclusively occupy coastal environments, although they are also found in fastflowing inland rivers. They are largely solitary animals and are today very difficult to approach in the wild. Like the small carnivores noted above, otters may however also have been an important source of meat, as well as hides. Butchery marks on the bones of otters at Cnoc Coig, Oronsay, Scotland (Mellars 1987) however suggest that these otters were taken principally for their pelts.

The densities of small mammal predators could have been very variable in different environments. Whilst the densities of small herbivores and omnivores are dependant on plant food and insects, the densities of predators would be largely dependant on the availability of their prey. Diets of small mammal predators are often quite varied however, the diet of red foxes includes beetles, earthworms, birds, fruit and carrion (Charles 1997) as well as small mammals. Badgers also eat similar foods, predominantly exploiting earthworms. In general terms, the densities of small carnivores would be highest in open woodlands with a rich soil litter, as these environments would support a rich ground layer vegetation with densest insect and small mammal fauna. In riverine environments, another small mammal, the beaver, would have been found. The bone remains of six or seven beavers, with cut marks, were recovered from Star Carr (Clark 1954; Fraser and King 1954; Charles 1997). Beaver could have been an important source of fat, since 30-40% of their body weight is fat, even in winter, as well as a source of meat, teeth and pelts. Beavers inhabit lakes, rivers and streams, especially near the coast. They largely exploit tree bark, buds of willow and birch and various aquatic herbs and plants. Keene (1976) reports that the optimum season for beaver exploitation would be the winter, when they are slow moving and in 38

coast, acorns (alongside game) were a major predictor of population densities.

PLANT FOODS It is noticeable that land mammals, and large land mammals

Unfortunately, plant foods are largely an 'invisible' component of diets. However, charred nut remains present something of an exception, and thus there is evidence for nut exploitation in Mesolithic northern England. Remains of charred hazelnuts have been recovered from four sites - Star Carr (Clark 1972), Blubberhouses Moor (Davies 1963), Thorpe Common (Morrison 1980) and Cass ny Hawin (Woodman 1987). Charred hazelnuts have also been found at twenty other British sites (Zvelebil 1994) and both charred hazelnuts and charred acorns have been recovered from Mount Sandel in Ireland (Woodman 1985b; WijngaardenBakker 1989). Nygaard (1987: 149) even noted that in South-western Norway 'there is hardly a site without some remains of charred nutshells'.

in particular, have in the past not only been considered the main resource in early interpretations of subsistence practices, but moreover were seen as the only resource of interest. Two events changed perspectives on large land mammals, and introduced plant foods, rarely ever recovered and practically invisible on archaeological sites, as a potentially major component of diets. The first event was the 'Man the Hunter' symposium in Chicago in 1966 (Kelly 1995). Here, a range of archaeological researchers discussed general principles of hunting and gathering societies. Lee's work amongst the !Kung (Lee 1968; 1969; Lee and deVore 1968) was particularly influential in a re-appraisal of hunting and gathering. Mongongo nuts, rather than game animals, appeared to be a major staple for these people (who inhabited a semi-arid region of southern Africa). It was thus concluded that the role of large animals in contrast to plant foods in hunter-gatherer diets had been much inflated.

The availability of nut resources is dictated by the distribution and density of nut producing trees. Both hazel (producing hazelnuts) and oak (producing acorns) would have been common trees in postglacial forests, although the distribution of the two tree types varies somewhat and changed throughout the period. Hazel was a major component of Early Mesolithic forests (Iversen 1973; Birks 1989). It would probably even have grown as forest stands (Bennett pers. comm.) prior to competition from more shadetolerant species. The spread of dense deciduous woodland in the lowlands from the Early to the Late Mesolithic would have restricted hazel distributions however. By the Late Mesolithic, hazel would still have been abundant in the upland zone, as a major component with birch of the upland forest, and as a scrub above the closed canopy woodland (Simmons 1996: 21) but would have been restricted to clearings or edges of climax woodland in the lowland zone (Simmons 1996; Keene 1981: 70). Oak, in contrast, arriving to Britain at the end of the Early Mesolithic, would have been a major component of these Late Mesolithic woodlands.

The second event, following on from this change in perspective, and from studies of the potential advantages of plant foods, such as Dimbleby's (1967) 'Plants and Archaeology', came Clarke's (1976) essay 'Mesolithic Europe: The Economic Basis'. Clarke (1976: 464) pointed out the large numbers of edible plant species which should have been available in the Mesolithic (200-450) and the biased nature of an archaeological record in which faunal remains are the primary evidence of subsistence. He also made a fundamental point - that it is not just biased evidence that influences our interpretation of Mesolithic subsistence but also the current academic and social climate. Furthermore, Clarke (1976) also noted that the dominance of microliths in Mesolithic assemblages, interpreted as barbs for arrows used to hunt large game, had been a major influence on the concept of the dominance of large game in Mesolithic diets. Clark countered this perspective by suggesting that microliths might have performed other functions as plant processing equipment (1976: 453-456).

Both types of nut are ripe in autumn but there are distinct differences in availability, collection and processing of the two sp~cies. Whilst hazel produces nuts fairly consistently, oak typically produce a good 'mast' crop only every 3-4 years (Keene 1981). There is also a lot of yearly variation in acorn production, Park (1942) illustrates that total crop failure is common, production is spatially very variable and only 2030% of oak trees bear fruit in any given year. The effects of competition form the many animals (such as squirrels or wild boar) which exploit nuts can also be important. Keene (1981: 70) for example notes that the Meskwaki of the North American Great Lakes collected nuts before they were ripe because of competition with animals. A further advantages of hazel is that hazel, unlike oak, can produce abundant nuts even when subject to heavy wildlife predation.

In view of this change of perspective, and although it is often difficult to determine with absolute certainty if 'native' plants were available in the Mesolithic (Mabey 1996), the discussion below demonstrates the range of possible plant foods available to Mesolithic populations, several of which have been recovered from archaeological sites in northern England, the British Isles as a whole, and in the rest of northern Europe.

Nuts The key element in arguments for a major role for plant foods in hunter-gatherer diets has been nut resources, and in Mesolithic Europe specifically hazelnuts and acorns. Nuts would have been readily available, with nut producing trees being typically abundant in the wooded environments which characterised the Mesolithic. Nuts are high in protein and fat, moreover in the highly seasonal environment of temperate north-west Europe they are a storable resource. In fact, nuts are even known to provide a staple resource for certain hunter-gatherer populations. Baumhoff (1963: 223) has shown that in the California province of the American West

As well as collection costs, the costs of processing are another factor to consider. On top of being less predictable as a resource, acorns also require more intensive processing than hazelnuts. Both types of nut require time-consuming collecting, and shelling (although hazelnuts are easier to shell if dried, Keene 1981: 72), however acorns must also be leached (in hot water or in cold if previously broken) for

37

several hours to remove tannic acid (Keene 1981: 75), so the costs of processing are particularly high.

Tubers are in general most abundant in damp environments. Terrestrial tubers are commonly found at stream edges and in damp woods, and in Britain are associated with alder woods (Rieley and Page 1990: 66; Mabey 1996). Aquatic tubers would have been abundant at the edges of ponds, lakes and slow-moving rivers. White water lily for example, growing in relatively deep water would have produced productive storage organs throughout the winter (Tilley 1979: 19; Price 1989; 50). In shallower water, tubers of water plantain, water parsnip, and clubrush would have been available, with sedge communities also providing edible roots (Clarke 1976). At the coast, wild carrot could be locally abundant, with two species of wild carrot, one common on cliffs and the other on dunes and grassy places near the sea (Soothill and Thomas 1987). Other species of plantains and edible roots such as sea parsnip would be found on the muddy margins of intertidal rivers. Tubers are also found in less damp environments, wild parsnip for example would have been found on chalky soils (Mabey 1996).

Paradoxically, though hazelnuts appear to be a more attractive resource, it is acorns, rather than hazelnuts, which are most commonly a staple food amongst historic native populations in the Great Lakes and throughout the Eastern United States (Keene 1981: 55) as well as California (Baumhoff 1963). However acorns may not have been the most important nut resource in the European Mesolithic. For one thing, American and European oak trees are a different species, with the latter not necessarily as suitable for exploitation (in fact acorns are rarely even fed to livestock in Britain because they are noted to give digestive problems). For another, the processing of acorns in the Eastern United States involves the use of mortars for which we have no evidence in Britain. A further factor is that it is, in any case, hazelnuts which are the most common charred nut remains on archaeological sites. The main reason why nuts figure so highly in arguments for the importance of plant resources in diets is that the energetic 'returns' on nut exploitation are relatively high (ranking for some authors in the range of, and even above, small game animals - Perlman 1980, see table 3.2). Most other plant foods are less obvious candidates for being a major food source, with greens or fruit for example, showing relatively less energetic returns. Some other plant foods, being less 'visible' than nuts, are however overlooked and might nonetheless have been important. Roots and tubers, for example, are a valuable source of carbohydrate, and can be important in winter when other resources are scarce (Keene 1981; Dimbleby 1967). In fact, it is often these types of resources which, though not necessarily a major contributor to overall diets are often the major determinant of population numbers (Casteel 1972).

Tubers are most abundant in late spring and autumn, with some species still potentially being important resources throughout the winter. Additionally many species, such as wild leek and wild onion in terrestrial environments, and white water lily in aquatic environments, also produce edible greens that can be used in the spring. In autumn, where tubers are concentrated, the time taken to gather and process this resource is the main limitation on exploitation rather than availability (Keene 1981: 85; Dimbleby 1967). Though often overlooked, tubers are important for certain ethnographic populations. Tubers were even dried for storage by the historic populations of the temperate forests of the Great Lakes (Keene 1981: 83) and also figured very highly in the diet of groups in arid environments such as in areas of Australia (Cane 1987). Hence, as a resource, they would have been abundant in damp, open environments in the Mesolithic, even in the winter months, could have provided a valuable source of carbohydrate, and would generally not have been difficult to process.

Roots and Tubers There is little archaeological evidence for the exploitation of tubers, which normally decompose rapidly. One unique exception in Britain is in Scotland, at Staosnaig on Islay, where fragments of charred tubers of an edible species of buttercup were recovered (N. Finlay pers. comm.). Indirect evidence may survive however. It has been suggested that perforated antler or bone mattocks found in northern England, (Wymer 1991: 24, Wickham-Jones 1994: 94) or stone picks in southern England (Palmer 1977: 184), may have been used for digging up roots and tubers. Certainly, macroscopic wear traces and damage on antler and bone mattocks are consistent with their use as heavy duty digging tools (Smith 1989).

Another resource which can easily be overlooked, yet can also be a source of carbohydrate available in winter months, are seeds.

Seeds Seeds, like tubers, are rarely preserved. Nonetheless, wetsieving of occupation levels in Scotland, at Morton in Scotland (Coles 1971) did reveal a few seeds typical of waste land (chickweed, fat-hen and com spurrey). Fat-hen seeds (goosefoot or pigweed) were also recovered at Mount Sandel in Ireland (Zvelebil 1994; Monk and Pals 1985). Fat-hen is however typically associated with later prehistoric and historic occupation areas, its densities being limited in the Mesolithic since it is largely a coloniser of disturbed ground (although densities could be increased by burning). Paradoxically, although the main evidence for seed exploitation in Mesolithic Britain comes from a species typical of dry areas, the best potential environment for seeds is actually in aquatic and damp environments such as lakes and slow-flowing rivers. In these environments the yield of Clubrush seeds, for example, outstrip by 30% that of most

Whilst archaeological records for the exploitation of tubers are limited, historical records do document the exploitation of native tubers in Britain. Pignut roots, found in open woodlands are noted to have been a delicacy for example. Mabey (1996: 288) notes a comment that 'you could dig up enough [pignut] to feed four people in half an hour'. Tubers were even important as a staple in times of famine. The roots of silverweed which would have grown in more open areas and waste or disturbed ground, were once eaten in upland regions and even dried and baked into a flour (Mabey 1996: 186) when wheat flour was scarce. 38

modem dry-land cereals (Phillipson 1966: 37). The seeds of many lake plants, such as yellow water lily and white water lily, are also known to be potentially important edible resources (Clarke 1976). In fact, yellow water lily seeds were recovered from excavations at Derravaragh in Ireland (Zvelebil 1994; Morrison 1980; Mitchell 1972). Moreover, at the coast other seeds would also have been available such as sea clubrush, growing in muddy inter-tidal rivers. Fat-hen seeds were an important resource for hunter-gatherer groups in the Great Lakes (Keene 1981). They were gathered in autumn with the best yields being after the first frost. Nonetheless although seeds are very productive (as noted above for clubrush), the costs of exploitation are very high. Seeds are time-consuming to remove from husks, and also need to be cooked to make them edible. Thus, significant seed exploitation is a possible strategy for hunter-gatherer groups to adopt where resources are very limited, or where population pressure is intense. However, the processing costs of seed exploitation tends to make seeds less attractive than other resources, and groups which are heavily dependant on seeds tend to only be found in arid environments (Wright 1994) where other resources are often scarce. Apart from nuts, tubers and seeds, most other plant resources could make only a minor contribution to diets in terms of energetic input. However, other plant resources often provide important nutrients, especially vitamins. As such, other plant resources such as greens and fruit might, for example, have dictated exploitation patterns in certain circumstances.

Greens Like tubers, there is limited evidence for the exploitation of greens (the edible leaves of undergrowth plants). There is no evidence for the exploitation of greens in Britain, but sorrel has been recovered from Mesolithic occupation levels at Agerod Vin southern Sweden (Zvelebil 1994; Larsson 1983; Goransson 1983). Many species are however known to have been eaten in the historical past, such as sorrel, cowparsnip, dock and cress (Mabey 1996). Even the common nettle proved useful as a subsistence resource in the Irish potato famine and in the second world war (Mabey 1996). Terrestrial greens would have been abundant in spring, alongside streams and in damp woods, although in general terms, greens would be available wherever sufficient openings existed in the forest canopy. Aquatic greens include edible watercress (Clarke 1976: 465) which can grow densely in lakes and in slow-flowing rivers, as well as marsh marigold, cress and sedges (Soothill and Thomas 1987). At the coast, wild cabbage, fennel and sea kale occur on sea cliffs, sea rocket and sea holly along the drift line, with wild celery in rivers and ditches near the sea. The most famous of coastal greens include some species of seaweed - such as dulse, Irish moss and laver - that continue to be eaten along the western coastline of Britain in the present (Mellars 1978: 379-380). The major 'cost' of exploiting greens is in collection, especially as these types of plants are often widely scattered. The leaves also often need to be boiled (although sorrel can be eaten raw and is eaten in salads today). Though greens might seem an attractive resource, the greens which we eat

today from supermarkets are very different from their wild ancestors, wild greens such as sorrel and chicory are usually described as 'bitter tasting' or 'sour', or as in the case of wood sorrel, are slightly toxic in large quantities (Mabey 1996: 96).

Fungi Like greens, many different species of fungi would have been available in Mesolithic forests, with most species of fungi preferring damp, dark environments. Edible fungi (such as puffballs, Bassett 1997) are generally collected in autumn, and tend to concentrate in the same locations each year as they grow from underground, thread-like mycelium. Fungi would also be a valuable source of additional vitamins and minerals, though again, like greens, of only minor energetic benefit.

Fruit There is no direct evidence for the exploitation of fruit in northern England. However wild pear or apple seeds have been recovered in Ireland at Mount Sandel (Zvelebil 1994; Woodman 1985b), raspberry seeds at Newferry (Zvelebil 1994; Woodman 1978b) and in Scotland, barren strawberry seeds at Lussa River (Zvelebil 1994; Mercer 1970). Though hardly a dietary staple, fruit may have been an important source of vitamins and minerals in late summer and autumn as well as a source of dietary variety. Fruit could generally either be available from shrubs (shrub fruit) or trees (tree fruit). Strawberries, on the other hand, would grow as an undergrowth plant. Most shrubs in open deciduous forest bear fruit, though not all of this fruit would be edible (Keene 1981: 76). Brambles may even form 70100% of the cover in forest clearings. Though closure of forest openings takes 7-10 years, dense stands of certain species can even survive for around 15 years (Keene 1981: 77). Crop failure is not common with shrub fruit, and the time of ripening is consistent. However, the collection of shrub fruit is time-consuming, and, as with nuts, predation by small animals and birds on shrub fruit can be heavy. In reality, the presence of these thorny fruit shrubs in open forests may have been more of a curse than a blessing since brambles would have made access and movement extremely difficult. Tree fruit provide an alternative fruit resource. These tend to be found in open woodlands, since fruit trees, such as wild cherry, are shade intolerant. The cost of exploitation of tree fruit is also high however (Keene 1981: 82) as fruit ripens unevenly, is difficult to access, and mature trees tend to be widely distributed. The availability of bush and tree fruit would have changed as forest types and distribution changed throughout the Mesolithic. Neither fruit shrubs nor trees are shade tolerant, and undergrowth plants supporting strawberries also need light conditions. The gradual closure of the lowland forest canopy would thus have restricted fruit distributions in the lowlands to gaps in the forest canopy caused by tree falls, or at the edges ofrivers and lakes. Fruit are sometimes collected by ethnographically documented populations for winter storage. Hawthorn fruit, growing in clearings and stream banks, were collected in autumn by Great Lakes populations and dried for the winter (Keene 1981). However in general terms, fruit are never a major resource, often being difficult to collect and in some

cases (particularly tree fruit) also unreliable with frequent crop failure.

OTHER USES OF PLANT RESOURCES

Although we are largely concerned with subsistence resources, it is worth noting that non-food plants may have been as important in influencing hunter-gatherer settlement systems as food plants in certain circumstances. Keene (1981) points out, for example, that of the 130 plants recorded by Yarnell (1964: 79-88) as economically important only 48 were used solely for food. Trees and bushes are perhaps the most obvious key plant resources. Although much of Mesolithic Europe was covered by forests, very specific trees and bushes with certain characteristics (such as strength or length of straight stem) would be required to make bows, projectile hafts, boats or canoes and shelters, and these specific requirements might only be met by trees in specific locations. Plants are also used to make traps and nets, shelters and means of transport (canoes such as that recovered at Friarton, in Scotland, Wymer 1991: 37) as well as containers and as a source of firewood. Certain plants can even perform very specialised functions, soft rush for example, can be dipped in fat to make lights, and soapwort can be used as a detergent (Mabey 1996).

One resource, strictly a 'fruit' although often classed as a nut, which might have been important is water chestnuts (Rowley-Conwy and Zvelebil 1989: 55). Water chestnuts have been recovered from a number of archaeological contexts in northern Europe, sites such as Agerod V (Zvelebil 1994; Larsson 1983; Goransson 1983) and Skateholm (Zvelebil 1994) in Sweden. Although present in the British Isles in the Early Holocene, wtaer chestnuts have not been recovered from Mesolithic sites. Like waterlilies they are found in lakes and ponds, like true nuts they could have been an important source of protein. Other edible plant resources also exist, such as tree sap, shoots and bark, although these parts are not known to be dietary staples. Tree sap is known to have been eaten by several North American hunter-gatherer groups, including those in the Great Lakes (Keene 1981: 87), however there is no historical records of the exploitation of sap in Britain. Lucas Bridges (1948: 304) comments that whilst tree sap is eaten by the Ona of Tierra del Fuego, only very small amounts were consumed as sap can be difficult to digest. Even tree shoots can be eaten if cut young and boiled (Dimbleby 1967), nevertheless, like tree sap, tree shoots were probably not palatable in any great quantity.

A further non-subsistence use of plants is as medicines. Several mainstream medicines include remedies derived from native British plants, including aspirin (from willow bark), cokhiare, used to treat gout (obtained from meadow saffron) and digoxim, used to treat heart attacks (obtained from foxgloves) (Mabey 1996). Those medicinal plants which were used in the Mesolithic may have included common comfrey amongst others. Comfrey is found by streams and rivers and its roots are usually exploited in the spring. Comfrey is also known as 'bone set', as it contains allantoin which promotes healing in connective tissue when applied (Mabey 1996: 307). It was eaten widely in the second world war but can cause liver damage in large quantities. Other remedies include common valeris, which can be used as a sedative and grows on the chalk downs as short plants, as well as on river banks and woodland clearings. Foxglove can also be used as a heart stimulant (although it can be lethal), and dandelion is a diuretic, which can be used to prevent gout. Lastly, hop grows on fens and river banks, it is recommended as an appetite suppressant, pain killer and sedative (Mabey 1996).

Even if plants are not edible they may still be an important part of exploitation strategies. In the same way that animals are often exploited for bones, antler and skins as well as for meat, plants may also be exploited for different reasons, with many plants having important 'non-subsistence uses'.

40

DISCUSSION

There would clearly have been an wide range of different plant resources available to Mesolithic populations, several of which could have been an important source of fat, protein and energy, as well as essential vitamins and minerals. Evidence for a range of plant resources has been recovered from archaeological sites. However, one problem with the archaeological evidence for plant exploitation is that whilst faunal remains may provide clear evidence of butchery, the presence of edible plants on a site is not necessarily evidence of their exploitation. Plants can be accidentally charred, and might have either grown near to a site or been brought to a site for non-subsistence purposes. A further limitation is that, even evidence of exploitation would not necessarily provide any indication of the relative importance of any species in particular or of plant resources in general. There are several reasons why the role of plant foods in diets may have been overemphasised by the results of the Man the Hunter symposium, and by Clarke's (1976) essay. The most obvious relates to the source of inspiration for the importance of plant foods. Many of the surviving (and studied) hunter-gatherers, specifically the !Kung, who formed the basis for ideas about typical subsistence practices, tended to live in the most arid regions of the world. These are the regions where plant foods have been found to be relatively more important in hunter-gatherer diets than in other environments (Kelly 1995), and because of this the wider role of plant foods is often overemphasised. Kelly thus points out that the 'generalisation that hunter-gatherers rely primarily on plant food is the result of differential ethnographic documentation' (Kelly 1995: 70). Equally, the factors governing the exploitation of different resources are now realised to be much more complex than was initially assumed. Those who supported the idea of a dominant role for plant foods focused particularly on the abundance and high yields of plant resources. However, the importance of plant foods in the subsistence strategies would be affected by several criteria, not only abundance, yields and nutritional value (whether calories, protein, fat or minerals) but also search time, ease of procurement and processing as well as location in relation to other resources (water or good areas for hunting game or fishing) and of course, palatability. Animal foods in general are much less time consuming to process than plant foods, and animals, being higher up the food chain than plants, provide a mix of complex nutrients more similar to our own than any plant resources (Kelly 1995), thus often being a preferred resource. Another point to consider is that, though the range of plants available to Mesolithic hunter-gatherers is large (Clark 1976) not all plants would have been edible. Dimbleby (1967) and Simmons (1996: 163), for example, propose that bracken was a major edible plant food resource, with Simmons even classing hazel and bracken as the most important plant foods (1996: 192). Bracken rhizomes can of course occur in large numbers in unforested and open forest habitats. However, Mabey does not include bracken as an edible plant and no records have been made of the use of bracken for food in Britain. Mabey (1996: 17) describes the young shoots being eaten as food in the Near East, but bracken is toxic to all

animals and carcinogenic if eaten in excess. Even where plants are strictly edible when leached or boiled, like acorns, many may be toxic without processing. The issue of the role of plant resources in Mesolithic diets has been one over which there has been little academic agreement. Several authors have maintained the importance of large land mammal resources in their interpretations. Smith (1990: 15) for example considers that subsistence practices in the Mesolithic might reasonably be taken to comprise 80% ungulate meat, with the rest being derived from small game, waterfowl and fish, as well as plant foods. Rowley-Conwy (1980: 189) suggests that diverse plant foods would have been scattered and 'unlikely often to have been as overwhelmingly important as Clarke makes out'. On the other hand plant resources are considered to be of prime importance by other researchers. Wymer (1991: 24) proposes that 'on the basis of ethnographic parallels, it would not be surprising if the diet of Mesolithic peoples consisted of a very high proportion of vegetable food'. Simmons (1996) accords an important role to many plant foods including not only hazelnuts but even bracken rhizomes. Zvelebil (1994: 58) proposes that the contribution of plant foods to the diet was probably greater than the 1520% estimate which is commonly used, and the use of plants more intensive than has been supposed. He makes a case for ' 'wild plant food husbandry; rather than the opportunistic use of plant foods, across Europe in the Mesolithic'. Interpretations of the importance of plant foods in Mesolithic subsistence are also affected to a great extent by preconceptions of the period, as much as by what very limited evidence exists for subsistence practices. Though the proportions of plant foods in diets are often estimated with confidence, these estimates vary widely, from 5% (Rozoy 1978 for France and Belgium) to 80% (Clarke 1976) with 15-20% being the most common estimate (Jochim 1976 for south-west Germany, Price 1978, for the Netherlands, Zvelebil 1981 for Southern Finland), as noted by Zvelebil (1994: 58). Explicit models have been used to determine the relative role of different resources, however these approaches are seriously limited by biases in ecological and ethnographic evidence used to construct them (see chapter four). Certain key characteristics clearly influence which resources are exploited, and how they are exploited, but specific regional and historical circumstances make precise determinations of subsistence difficult (discussed below). Whatever their overall contribution to diet, plant resources may nonetheless have been important however because they are an abundant, predictable and an easily managed resource. Plant resources may provide reliable resources in very variable environments, and also may be vital as a backup resource in poor seasons, either through the exploitation of surviving tubers or through storage, for example of nuts. In fact, plant resources may also exert an influence on settlement systems which is not necessarily proportionate to their subsistence contribution. Known hunter-gatherer societies commonly operate on the basis of a division of labour, with women tending to be more responsible for the collection of reliable resources such as the gathering of plant foods and collection of fish and small game, and men being concerned with hunting larger prey (Jochim 1988). Although

the extent to which women are involved in hunting either in communal hunts or in the routine hunting or trapping or snaring of small mammals does vary, the resource exploitation system is nonetheless planned to accommodate the procurement of both 'male' and 'female' resources. It is these 'female' resources, predominantly plant foods, which, being heavy and bulky, have the most influence on the location of long term occupation sites ('base camps'). Thus the distribution of plant resources (largely in damp environments and open woodland) is crucial in that it is these resources, whatever their contribution in calorific terms, which often determine the location of these larger and longer term occupation sites.

MARINE RESOURCES Academic attention has been focused on marine resources following the recognition of the importance of these resources to sedentary and semi-sedentary hunter-gatherers, such as groups on the Northwest coast of North America. For these groups, the availability of year-round resources at the coast, particularly marine resources such as sea mammals and fish, resources at the shoreline such as shellfish, and migratory resources, especially salmon, appeared to be the key factor sustaining sedentary communities and allowing the developments of elements of 'social complexity' such as storage, distinct territories and cemeteries (Price and Brown 1985; Keeley 1988; Rowley-Conwy 1986).

Apart from their direct relevance to discussions about Mesolithic subsistence practices, suggestions that plant foods played a major role in subsistence have also had the indirect effect of broadening the scope of discussions of subsistence. One other major group of resources, marine resources, which in all but select locations are also invisible in the archaeological record, subsequently came to take over from plant foods as the 'crucial but often overlooked resource' in the 1980s. Marine resources (or coastal resources in general) began to be seen as the key not only to subsistence practices of coastal groups but also to the important social changes taking place in the Mesolithic.

As discussed in chapter one, the main focus of our discussion is on inland groups. However, a consideration of coastal resources is nonetheless important, partly because these resources may have been an element which was included in the seasonal exploitation schedule of inland groups, as well as because some resources which are typically considered as coastal resources, such as salmon, are also exploited inland, and equally because models of subsistence and settlement which have been built around the exploitation of coastal resources provide a key contrast to models proposed for inland settlement systems (discussed in chapter four). We shall start with the largest and most obvious marine resources available to Mesolithic populations, that is, large sea mammals - such as whales, porpoises, dolphins and seals.

4.2

LARGE SEA MAMMALS

Whales (such as sperm whales or rorquals) are the largest sea mammals which might have been exploited in the Mesolithic. Whale bones were recovered in Scotland from Caisteal nan Gillean midden on Oronsay in the last century, and at Priory Midden in more recent excavations (Smith 1990: 150; Mellars 1987). McCormick and Buckland (1997) also note the remains of stranded whales at the Forth of Firth dating to between the mid-eighth and mid-seventh millennium bp, associated with red deer antler mattocks which may have been used to dismember the carcasses. Different species of whale are present off the coast of Britain at different times depending on their migration patterns. Although it is not impossible for whales to have been hunted at sea using boats, this would have been an extremely difficult and treacherous operation. Vorren and Manker note that the coastal Sarni in V arangerfjord did hunt small whales driven up to shore by killer whales, by using spears and pointed sticks from boats and 'might get up to ten small whales this way' (Vorren and Manker 1962: 59, cited in Renouf 1989). In general, for these people' whaling was a rather ... fortuitous business and mostly consisted of keeping an eye open for the carcasses of dead whales which washed ashore' (Vorren and Manker, as above). Whilst not a predictable resource, stranded whales would nevertheless have been an extremely valuable resource, which because of their shear size could feed a group for a long time. The opportunistic exploitation of the fat of beached whales could have been important particularly in times when only lean protein was available (the winter 'protein metabolism problem' described by Speth and Spielman 1983). Lucas Bridges (1948: 313) comments that for the Ona of Tierra del Fuego 'a whiff of a whale from leagues away was their intimation of this vast treasure of food; and they never wasted time in hastening to the scene'. At certain locations opportune, if not predictable, beachings might have been common. At Gressbaken in V arangerfjord (in the far north of Norway and so not within the map area of figure 3.1) a large number of whale species were represented in the Mesolithic faunal remains. It may be no coincidence that the coast at this point is conducive to whale strandings, with shallow waters and a gradually sloping foreshore that would distort echo signals (Renouf 1989: 210). Dolphins, another large sea mammal resource, like whales, would also have been difficult to exploit at sea, but unlike whales are not susceptible to beachings. Remains of dolphin or porpoise were recovered from Cnoc Coig midden on Oronsay (Mellars 1978; 1987), though it is unclear if these animals were hunted or isolated examples of opportunistic exploitation of dead animals washed ashore. Seal bones were also recovered from the Oronsay middens. In fact, remains of grey seal, at least nine or ten individuals including pups and adults, make up about 60% of the fauna recovered from this site. Seals, like whales and dolphins, are also difficult to exploit at sea, but unlike whales or dolphins they do come to shore at predictable times and locations where they would have been much easier to exploit. In the

breeding season (September to December), and in summer when onshore with young, large concentrations of animals would be present in coastal locations. Two main species of seals would have been available, grey (or Atlantic) seals and common seals. Grey seals collect on rocks in large colonies where they can still be seen all round the British coast except in the South. The smaller common seals, still found off the east and north-west coast of Britain today, are more likely to be found in small family groups in sheltered bays. Another difference in the possible exploitation of these two species is that whilst grey seal pups would have been on shore for two months before going to sea, common seal pups go to sea almost immediately (making them more difficult prey). Though it is generally assumed that seals were caught at the coast, they might potentially also have been caught up-river. In Norway, seals move up-river in autumn following abundant salmon runs (Renouf 1989: 34) and in historic times they were caught in these locations in nets. Seals could thus provide an important and potentially storable source of meat and fat as well as hides. Unlike large land mammals, seals are much less susceptible to overhunting, and even winter seal meat is very fatty (whilst terrestrial mammals are normally very lean in winter months). It has been suggested that seal hides were particularly important in Ireland, as red deer did not colonise the island until late in the Mesolithic, and the main other large terrestrial mammal - wild boar - does not have hides suitable for use (Wijngaarden-Bakker 1989). Apart from direct evidence for seal bones on archaeological sites, there is also possible indirect evidence for the hunting of large sea mammals. Bone or antler 'harpoons' which might have been used for hunting seals, as well as fish, are characteristic of the British Early Mesolithic. Several bone harpoons were recovered from Star Carr (Clark 1954). Marine environments also support other, potentially equally as important resources, including marine fish. As is the case with whales and dolphins, a lack of knowledge of the seafaring capacities of Mesolithic populations and their precise means of exploiting different resources limits interpretations of the potential importance of marine fish.

MARINE FISH

Salmon The importance of seasonal salmon 'runs' in providing abundant storable food for semi-sedentary and sedentary Northwest coast populations in North America has attracted attention to the potential importance of salmon for Mesolithic populations. As noted above, fish bones are in general rare on Mesolithic sites, but in Britain those of salmon are particularly scarce. Bones of either salmon or trout (since it is difficult to distinguish the two) were however the dominant species among the 1800 fish remains at Mount Sandel in Ireland (Wijngaarden-Bakker 1989).

Fish bones are not often preserved on archaeological sites, and methods of excavation are in any case rarely thorough enough for the recovery of fish bones. Like other marine resources (such as whale, dolphin and seals) the main evidence for the exploitation of fish in Mesolithic Britain comes from recently excavated Scottish coastal shell middens. Often one fish species dominates others at these sites. Over 90% of the identifiable fish bones at Morton were of cod, for example, (with haddock bones also recovered, Coles 1971). Mellars (1978; 1987) even suggests that the main reason for the occupation of the midden sites of Oronsay would have been marine fish, rather than shellfish. Whilst saithe make up 95% of the fish remains from Cnoc Coig (Mellars 1978; 1987), it is suggested that the shellfish recovered at the site may actually have been used as bait for catching saithe in inshore waters. Sea fish were also well represented at Risga, including skate, grey mullet and haddock (Wymer 1991: 37).

The relatively scant evidence for salmon exploitation contrasts markedly with their potential availability. Salmon would have been found in all British rivers in the Mesolithic with the largest salmon runs tending to be on the larges~ rivers. There is a long history of salmon exploitation in Britain in general and in northern England in particular. In fact, in 1806 almost all the main rivers in England were prime locations for fishing for salmon - notably the Thames, Severn, Mersey, Trent, Medway, Exe, Usk, and Wye (see figure 1.1). Even a single angler could catch a phenomenal weight of salmon on these rivers - one man, Robert Pashley, hooked 9,800 salmon from the Wye between 1906-1951 (Netboy 1968). In fact, salmon were so abundant in the rivers of northern England that they have until recently been seen as a peasants' food - Thomas Bewick, a wood engraver in the 1760s in Newcastle, had an article inserted into each contract of apprenticeship that apprentices should not be forced to eat salmon more than twice a week, the same bargain being made with servants (Netboy 1968).

Several species of marine fish, such as cod, whiting, haddock, pollack, and saithe, could potentially have been caught not far from the shore using boats in autumn and winter. Since they are available in winter, marine fish could potentially provide an important 'poor season' resource. In fact Casteel (1972) has demonstrated that for the Chipewyan, fish availability was the main determinant of population numbers. Furthermore, fish are also potentially important as a potentially storable resource. Amongst marine fish, cod and saithe could be dried directly, but more oily fish would need to be salted to be stored.

Salmon would be available at times of salmon 'runs' when fish migrate from the Atlantic and North Sea to the sources of major rivers to breed. In Britain today, the west coast rivers have mostly only summer-autumn runs, whilst in the east coast there are spring runs as well (Netboy 1968). This might have been a significant contrast in the Mesolithic, although it is uncertain if salmon behaviour was similar to that today, particularly given early postglacial changes in river flows, and rising sea levels. For the Tolewa of Oregon, Deith noted that groups based on rivers with a single salmon run were more dependant on acorn-gathering and hunting than those groups based on rivers with two salmon runs (Deith 1989: 74).

The 'returns' on sea fish exploitation are clearly heavily dependant on the technology used (with large nets for example making the exploitation of large numbers of fish possible). Evidence for the technology of fish exploitation in Britain includes a dugout canoe found at Friarton, in Scotland. This canoe, apparently capable of offshore transport, was about four and half metres long and a metre wide and made from pine wood (Wymer 1991: 37). Whilst bone or antler harpoons may have been used to spear fish, there is only one certain fish hook known from the British Mesolithic, this was recovered from excavations on Oronsay (Mellars 1978; 1987; Wymer 1991: 37). In Scandinavia however there is much more extensive evidence for fishing techniques, with waterlogged sites yielding nets, canoes, paddles (such as those at Tybrind Vig, Andersen 1985) and even eel traps (such as at Agerod V, Larsson 1985) , although it is difficult to assess how relevant the suite of techniques used in Scandinavia are to the British Isles.

In general terms, where spring runs exist, early fish are in prime condition in the spring, schools of smaller 'grisle' appear later, with runs of full grown salmon peaking in July and tapering off in hot August days. Salmon runs resume in September to October (spawning November to January) and salmon arrive upstream by October-November in poor condition (at this time lower pools are empty). Although salmon runs rarely fail, the timing and productivity can vary markedly. Rowley-Conwy and Zvelebil (1989) illustrate the present variability of up-river salmon runs on the Clwyd and the Dee as a potential model for variability in the Mesolithic. Variable salmon runs are determined by climatic variability, mild winters for example will result in more spring fish than colder conditions, with salmon reaching the upper reaches of rivers earlier (February-March), and after a cold winter salmon will not reach upstream until April-May.

One particular fish has drawn the greatest amount of archaeological attention, that is salmon. Although perhaps more appropriately seen as a riverine, rather than a marine fish, salmon spend most of their lives in the marine environment. They would however have been predominantly exploited at the riverine stage of their life cycle, when migrating up-river in annual salmon 'runs'. Salmon are a key element of discussion of possible sedentary societies, having been a major resource for American Northwest coast huntergatherer groups.

44

salmon, eels spend most of their life in freshwater lakes and streams, only upon reaching adulthood do they migrate to the sea to spawn and die (Wijngaarden-Bakker 1989: 129). Eels might have been caught using pronged forks (eel forks) where they were visible in clear lake waters. Examples of such potential eel forks exist in Germany (Wundsch 1962 in Wijngaarden-Bakker 1989). They could however also have been caught in muddy waters by using nets or baited lines. At Agerod V, fragments of a large wicker cage is likely to have been an eel trap (Larsson 1983).

The main reason why salmon productivity in northern England today is low is due to both over-fishing and industrial pollution. Although the effects of over-fishing were already being felt in 1215, when legislation in the Magna Carta ensured that weirs were removed and salmon had free passage to spawning grounds (Netboy 1968), it was industrial pollutants and canalisation which sealed the fate of British salmon, particularly for the productive rivers of northern England. The major decline of the fish stocks occurred last century. By the time of the report on the salmon fisheries of 1869, of those English rivers that should have been productive, only a quarter produced any salmon (Netboy 1968).

Although trout and eels might also have been important resources, salmon appears to have been clearly the most abundant migratory fish resource in the Mesolithic. Nonetheless, the marine resource that has perhaps most characterised the Mesolithic however, regardless of its productivity, as noted in chapter one, has been shellfish.

As a resource salmon are potentially important because of the very large quantities of protein that can be collected in a very short time. Salmon runs in the Dee and the Don supplied London with over 700,000lb of salmon in 1817 for example. Their overall importance as a potential dietary staple in the Mesolithic however depends on whether they were collected to be stored during salmon 'runs' or simply exploited opportunistically to provide a seasonal resource.

Shellfish Varying opinions have been voiced for the importance of shellfish in prehistoric diets, from being of 'minor dietary significance' to being a 'major food supply' (Bailey 1975: 45) - a discussion not dissimilar to that over the relative contributions of plant and animal resources.

Indirect evidence has been claimed as support for the role of stored salmon in subsistence practices at Mount Sandel in Ireland. Woodman (1978; 1985a) and Wijngaarden-Bakker (1989) suggest, on the basis of evidence for occupation at several different seasons (the presence of migratory fish species, the age of hunted wild boar and the presence of burnt hazelnuts), and on the basis of heavily built hut foundations, that Mount Sandel was occupied for most of the year, with winter occupation being dependant on stored salmon. Posthole arrangements at the site are interpreted as possible fish drying or storage racks, and several large pits may also have been used for storage (Woodman 1985a). Mount Sandel is ideally placed to exploit migratory salmon resources, and is close to other terrestrial and marine resources. The question of potential salmon exploitation is a difficult one however, since neither the presence of pits and hut floors, nor evidence for occupation at different times, provide significant evidence for either the intensive exploitation of salmon stocks or for permanent occupation.

There is evidence for the exploitation of shellfish at selected coastal sites in the British Isles. At the Scottish coastal site of Morton, Fife, cockle and Baltic tellin were the most common shellfish in the midden (Coles 1971; Smith 1990). At Oronsay the middens were largely dominated by limpet shells (Mellars 1978; 1987) and periwinkles and limpets with some crabs were recovered in the midden at Culver Well, Portland, Dorset (Wymer 1991: 35). Remains of crab were also recovered from Cnoc Coig midden on Oronsay (Mellars 1978). Crabs may not necessarily have been an immediate resource but, like shellfish at the Oronsay middens, may have been used as bait to catch marine fish. Shellfish in general would have been most common on rocky shores, alongside other shoreline resources such as crabs, and a range of plant resources (discussed earlier). Shellfish would have been available year-round, but the condition and 'meat weight' do vary. For example, in Norway, shellfish are avoided when spawning, and are preferred in the Winter, around Christmas (Renouf 1989: 33).

However salmon is not the only riverine fish that could have been exploited. Other riverine fish include trout and eels. Sea trout also have optimal catches in the summer, but not with the densities of salmon 'runs'. In Norway, sea trout ascend the rivers later than salmon, usually about the end of July (Renouf 1989: 34). In contrast, brown trout do not migrate and are territorial. Brown trout can however reach high densities in Irish rivers today (Wijngaarden-Bakker 1989) and trout appear to have been one of the resources exploited at Lough Boora in Central Ireland, where a salmonid vertebrae could be provisionally attributed to brown trout.

The location and methods of exploitation of shellfish, as well as the optimum timing for exploitation, would vary between species. Inter-tidal species can be collected by hand, whereas those species that inhabit deeper water have to be collected using tools used from a boat (Renouf 1989: 33). Cockles prefers sandy environments and can be collected on beaches at low tide, whilst the periwinkle is tolerant of a wide range of conditions, including algae covered rocks, small stones, soft mud and occasionally sand (Renouf 1989: 32). The time at which species spawn also varies in different regions with cockles in Wales spawning earlier than those in Scotland (Rowley-Conwy 1984, after Hancock and Simpson 1962: 38; Chambers and Milne 1979). Rowley-Conwy (1984) notes that the optimum season for the exploitation of oysters coincides with a low point in the availability of other coastal resources in Denmark. Oysters are in good condition in February to April, and most accessible because of spring

Eels could also have been potentially important as a seasonal resource. Eel bones were recovered in Scotland from Risga (Wymer 1991: 37), and in Ireland at Mount Sandel and Lough Boora (Wijngaarden-Bakker 1989). In fact, of 200 identifiable fish remains at Lough Boora in Central Ireland, 77% came from eel. Eel runs are confined to the autumn months, with a maximum occurring in October. Unlike 45

tides, he thus suggests that they plugged a 'gap' in resource availability for the coastal Ertebjljlle (1984: 306). The low overall calorific value of oysters does however argue against their role as a key seasonal staple, at least unless supplemented by other resources such as tubers or even terrestrial game.

MIGRATORY BIRDS

Seabirds can provide an abundant and predictable resource at coastal sites when nesting. At Morton, in Scotland, eleven species of birds were identified among the 217 bird bones recovered (Coles 1971; Smith 1990: 145), the most common being guillemot, gannet, cormorant and razorbill - birds which nest on rocky cliffs in the summer. Over thirty species of bird have also been identified from the Oronsay sites (Mellars 1978: 379) and again many would typically nest on cliff sites. Even quite large migratory birds may have been available. Bones of the now extinct great auk also were recovered from the Casteill-nan-Gillean midden on Oronsay (Clark 1948), and also from Risga (Wymer 1991). The great auk would have been a large bird and relatively easy to catch (which contributed to its extinction). Although a bird typically of shorelines in northern seas, remains of great auk have been found as far south as the Mesolithic site of Teviec in France (Clark 1948) and thus this species might well have been exploited at now submerged coastal sites in northern England.

Shellfish are frequently stored by Northwest coast huntergatherer groups. However in Europe shellfish are smaller than in north America (with the exception of oysters) and winter storms do not restrict access to the shore, so storage appears less advantageous (Deith 1983). Shellfish were probably most important as a source of minerals rather than for their calorific value, on shellfish alone 700 oysters or 1400 cockles would only provide for one person for one day (Bailey 1978). Shellfish might also have added some variety to diets. Deith (1989: 73) notes that northern Northwest coast populations exploited a wide range of shellfish which were often small and had little food value, largely because their diet was dominated by salmon, and shellfish added dietary variety. The relative importance of shellfish economies may have varied throughout the Mesolithic. It has been argued that shellfish may have become relatively more important through the Mesolithic of Cantabrian Spain (Clark 1983; Straus and Clark 1983), an argument which would tie in with a progressive diversification of the Mesolithic subsistence base (discussed in chapter one). In several Mesolithic shell middens in this region, both a decrease through time in size of limpet shells and an increasing representation of less easily accessible molluscan species are recorded (Bailey 1983: 162). The issue of progressive intensification is a difficult one however, as shellfish can be shown to be very susceptible to slight changes in ocean currents or salinity (N. Milner pers. comm)

Despite the pressures of industrialisation, pollution and impinging human settlement, major concentrations of migrating birds and of seabird colonies still collect at specific points along the coast in northern England today. Examples include several seabird 'colonies', such as over 1,000 breeding pairs of puffins, guillemots and kittiwakes at the Fame Island in the far north-east of northern England, and several sites for waders (such as the Dee estuary and the Humber estuary) where well over 25,000 individuals of any single species can be recorded (Soothill and Thomas 1987). The concentrations of these birds would be likely to have been much greater in the Mesolithic. Concentrations of coastal birds were an important resource for some recorded hunter-gatherer populations. Although almost impossible to catch at other times, hunter-gatherer groups often exploit coastal birds when they are nesting as they can be clubbed or snared with a noose. Lucas-Bridges (1948: 332) described three different methods which the Ona of Tierra del Fuego use to catch cormorants when they were nesting on cliffs. Practised hunters could catch hundreds of birds at their nesting sites while they were asleep. At Mesolithic coastal sites in Denmark, similar specialised procurement of migratory birds appears to have been practised at specific seasons (such as whooper swan at Havnj/.1in winter, guillemots and gannets at 0lby Lyng in the same season) as well as more opportunistic exploitation (such as of mute swan and mallard probably exploited in the summer at Mullerup) (Grigson 1989).

One further coastal, though not marine resource, which is often overlooked, but which, like salmon or seals may have provided an important seasonally abundant resource, are migratory birds.

Although apparently a 'small package' resource, planned (or 'logistical') exploitation of coastal birds could provide an abundant and potentially storable resource. However, many other coastal resources such as salmon, trout, eels or seals for example, also have the same potential for exploitation and storage. Determining what role different coastal resources actually played, is clearly a difficult matter.

48

lakeside situations, and riverine fish. There are however a whole suite of resources available in inland environments which are often overlooked, namely small mammals, and lake fish and birds.

THE ROLE OF DIFFERENT COASTAL RESOURCES

The first issue in comparing the advantages of different coastal resources is that the 'returns' of each resource varies greatly with the method of exploitation. Many fish for example, can be exploited more efficiently using nets, rather than line fishing. Equally, the usefulness of large marine mammal resources, such as seals, depends on how effectively they could be hunted. Unfortunately, though there is some archaeological evidence for different methods of exploiting marine resources, such as canoes, fish hooks or even nets, this evidence is insufficient to build up a clear picture of exploitation strategies. Whether resources were exploited intensively also depends on more than just available technology. What determines the subsistence strategy of any hunter-gatherer group, including whether resources are stored, are such factors as the structure of coastal environments (Rowley-Conwy and Zvelebil 1989), and also the history of subsistence choices taken by any population. Lourandos (1997) for example, notes that though abundant marine fish were available to hunter-gatherer populations in Tasmania, they were not exploited for cultural reasons. Another factor affecting whether resources are systematically stored is that storage often leads to sedentism, which is not an easy step for hunter-gatherers in coastal environments, since sedentism generates its own problems, such as localised reductions in large terrestrial mammal densities (Kelly 1995) and problems of diseases amongst large permanent populations. The characteristics of coastal resources are of some relevance for predominantly inland populations. Coastal populations in northern England might have remained at the same occupation sites throughout the year, through exploiting both the abundance and storability of coastal resources as well as the winter availability of resources such as fish and shellfish. If this were the case, then given the dense populations and territorial nature of present sedentary coastal groups (Keeley 1988), we might reasonably expect coastal resources to have been largely unavailable (except by trade) to inland groups. If, on the other hand, settlement systems at the coast were more mobile or more fluid, then predominantly inland groups might exploit coastal resources as one element of a seasonal exploitation system, much like the Ona of Tierra del Fuego exploited coastal resources for limited period in winter (Bridges 1948; Borrero 1997). As discussed in chapter one, it seems unlikely, given the distribution of raw materials, that inland groups, such as those exploiting the Pennines, would have a subsistence strategy which involved a substantial coastal component. Certainly it is assumed here that marine resources did not constitute a substantial component of the diet of inland groups. The question of potentially sedentary coastal populations, and the use of coastal resources is nonetheless an important one which is addressed in more detail in terms of models of coastal populations described in chapter four. The main resources of interest to us are however those in inland environments. Large land mammals and plant foods in open woodland environments have already been discussed, as have plant foods concentrating in damp and river and

47

del Fuego for example caught ducks individually using a captive duck as a decoy to lure other ducks near enough to kill (Lucas Bridges 1948: 97).

LAKE FISH AND INLAND BIRDS

At inland lakes, lake and river fish resources themselves would have been potentially important resources. As previously noted, fish remains are only rarely recovered. No fish remains were recovered from Star Carr, although the presence of birds which depend on fish, red breasted merganser, red throated diver and great crested grebe make a complete absence of fish in Lake Pickering unlikely.

Though it is coastal environments which are often considered valuable ecotones (regions crossing different ecosystems), inland lakes, and also inland rivers, would clearly have supported a wide variety of aquatic resources. Fish, as well as waterfowl at lakes, and beaver or otter in rivers, would have been available alongside plant resources with edible tubers and seeds, with terrestrial mammals also concentrating near more open woodland at riverine and lake edges. Schadla-Hall (pers. comm) even suggests that lake resources may have allowed permanent occupation within the area of the Vale of Pickering in the Early Mesolithic.

A number of different species of lakeside fish can be especially abundant in small or shallow lakes. Keene (1976) notes that the productivity of inland lakes of is a function of the amount of fertile shallow water and that small lakes tend to have a higher yield per unit area than large lakes. Lake fish could be caught be pole or line as well as by nets. Grigson (1989) notes that in Denmark, most inland sites on bogs or at lake edges show evidence of exploitation of pike. Fish resources in inland lakes would also attract wildfowl, such as ducks.

Inland open woodlands would also have been environments which supported a mix of plant resources and game animals, though not the variety supported by lakes and rivers. Some bird species, particularly small song birds and capercaillie would also have occupied open woodland. Capercaillie, a native of pine woods was amongst the faunal remains at Ertebjljlle (Rowley-Conwy 1983, Clark 1948). Pigeons would also be found in open woodland, with pigeon numbers and migrations patterns dependant on woodland nut production (Keene 1976). Pigeon is the main species of bird recovered at Lough Boora in Central Ireland and is present in 95% of the bird samples, with the jay coming in second place (Wijngaarden-Bakker 1989).

Swans and geese would be one of the largest 'package' bird resources, with ducks providing another potentially abundant bird resource at inland lakes and rivers. There are no records of exploitation of large water birds in Britain but a third of the bird bones at Mount Sandel in Ireland consisted of ducks and divers (Wijngaarden-Bakker 1989). Bird remains at Thatcham were also mostly of waterfowl, including mallard, teal or garganey, goldeneye and possible smew (Coles 1971; Smith 1990: 122). Both goldeneye and smew winter in Britain (breeding in northern Scandinavia and Britain). At Star Carr, a number of birds were recovered including water birds such as red-breasted merganser, red-throated diver, great crested grebe, little grebe, and duck as well as lapwing, crane and white stork (Clark 1954). Crane was also identified at Thatcham (Coles 1971; Smith 1990: 122)

Other birds, such as grouse, would be found in higher areas. Remains of both grouse and ptarmigan were recovered at Lough Boora, central Ireland (Wijngaarden-Bakker 1989). Birds of prey are also found in these open environments. Some of these birds of prey, such as eagles, might have provided a source of good feathers for fletching arrows (Clark 1948: 128), a possibility which is reinforced by the Scandinavia site of 0gaarde which appears to have been a site for the specialised procurement of eagles.

Some water bird species are year-round occupants, but many migrate, providing large concentrations of resources at certain seasons. Furthermore, migration patterns are usually quite consistent. Whooper swans, for example, breed in swamps and marshy tundra in northernmost Scandinavia and Russia, but over-winter in Britain and in areas near the North Sea. The faunal assemblage at Aggersund suggests that was a special purpose camp for procuring swans when they gathered there (Rowley-Conwy 1983 after M!Zlhl1978) with all the 257 bird bones being of whooper swan (Grigson 1989: 66). Keene (1981) notes that in the Great Lakes migrating duck would also occur in large numbers, with flocks of up to 26-50,000 along their migration routes (Jochim 1976: 117). The Ertebjljlle site of Sjljlager appears to have been a similar special-purpose camp concentrating on ducks (RowleyConwy 1983 after Skaarup 1973: 77).

Even in inland woodlands, there are far more resources than a brief survey of interpretations of subsistence practices might conclude. Other than large mammals and plants, small mammals, fish and birds might have been important resources. No one resource seems obviously more attractive to hunter-gatherer populations than any other. Hence, defining which resources may have been particularly important and how any resource fitted into an exploitation strategy is a difficult and complex issue.

The optimum time for exploiting water birds would be at the annual moult, as many species are unable to fly for a period of three to five weeks (Keene 1981: 118), when they could be caught in large numbers especially with the use of nets. Also of importance are small water birds, which, although thought of as small package resource, can be an abundant resource with special purpose exploitation, such as the use of nets. Other less 'logistically organised' means of capture are recorded in recent hunter-gatherers. The Y ahgan of Tierra 48

WHAT WAS THE ROLE OF DIFFERENT RESOURCES? Having considered the range of different resources available to Mesolithic populations, and the evidence for their exploitation, it is clear that many different resources could potentially have been a dominant or staple food. However, no one resource, or group of resources, is obviously 'the' key element in subsistence patterns. In the forested environment of the inland Mesolithic, large land mammals would presumable have been available and a much-prized resource, but, following a consideration of the range of other available resources, it is no longer self-evident that the meat from large game should be a dietary staple. The issue of a 'staple' resource is itself a problematic one, since it is clear that certain resources may take on a major role in survival (resources noted to be available at the 'poor season' for example) whilst not contributing the largest proportion in overall terms. Equally, certain resources may determine settlement systems because they are highly focused and abundant for a short interval, such as salmon 'runs', or because they are considered essential, such as medicinal plants for example, or even because they have a symbolic as well as practical importance, perhaps the furs of certain small mammals, whilst they may neither be represented in the archaeological record, nor ever have been a major staple. Mesolithic populations appear to have exploited a wide range of different resources. Taken as a whole, the evidence from northern Europe suggests the potential inland exploitation of a wide range of plant, large and small mammal, fish and bird resources. The precise importance or role of any particular resource in northern England is difficult to determine however, especially given the range of different resources considered above. There is very little direct archaeological evidence for the exploitation of food resources within northern England itself, and the evidence in neighbouring regions is of questionable relevance. We can make very general observations of the likely ecology and availability of different resources in the past, but moving from these general observations to comparing different resources in terms of the benefits or costs of exploitation either in general terms, or at different seasons, and to generating a model of subsistence, is problematic for a number of reasons. The most immediate problem is that any model depends on isolating important characteristics of resources, which can be compared to assess which resource is the most attractive at any time. Several potentially important characteristics have been discussed. These might include the season at which resources are available or easiest to exploit, the yields in terms of fat, protein or energy, and other characteristics such as how predictable they are, whether they are susceptible to over-exploitation, and whether they could provide potentially storable resources. It is already clear that no single characteristic is clearly the most important in defining how attractive any resource may have been. Incorporating the different characteristics of different resources into a model is not straightforward. For one thing,

different characteristics may be important in different circumstances. Large land mammals, for example, traditionally seen as the mainstay of Mesolithic economies, may be attractive because of the large 'package' of protein and fat which they provide, however they are a very unpredictable resource, which are difficult to capture. If it is important to bring home food, hunter-gatherers may opt instead to exploit more ubiquitous smaller game (Mithen 1987; 1989; 1990) or perhaps even very reliable resources which may be time-consuming to exploit such as nuts. Equally, resources such as hedgehogs or beaver, whilst only small 'packages' and possibly difficult to catch, may have been preferred in the lean winter season as they provide a source of fat. A further complication is that, as was clear in the discussion above, different plant or animal species, which appear to be similar, may have very different characteristics. Some waterbirds are migratory, and so would be available as abundant concentrated resources at certain seasons, being not only a potential source of short term food but also a potential source of storable food for the winter months, whereas nonmigratory water birds would have very different characteristics as a resource, being available year-round, and potentially important in winter, but rarely in large numbers. It would be almost impossible to build a model based on the availability of every individual different species of plant and animal, whereas combining resources into groups would confuse different characteristics. Perhaps the most serious limitation is that the preferences which hunter-gatherers exert over resources may not be predictable, but may be defined by particular historical or cultural circumstances or particular local or regional strategies. A pressure on available resources might perhaps encourage resources which are time-consuming or difficult to collect and process, and thus rarely exploited, to be included within diets for example. Subsistence strategies may even be governed by motivations which are not understandable today. As previously mentioned, Lourandos (1997) notes that Tasmanian population avoided fish, even though there were an abundant and potentially productive resource. Even if key characteristics could be isolated, the information may not be available to reliably compare different resources. Since our knowledge of Mesolithic environments is restricted, quoted densities or distributions of available resources are often guesswork. Large mammal behaviour in the past may have been very different from that recorded today (which is in any case often very variable). Even immobile resources are often difficult to quantify, Perlman (1980) for example suggests that the present returns on collecting acorns are generally about 48,000-60,000 kilocalories per hour, whereas Rowley-Conwy (1984), using a different source, suggests much lower returns of 18,00028,000 kilocalories. Without knowing methods of exploitation, it is equally difficult to assess how 'attractive' any resource may be. The main limitations to reconstructing Mesolithic subsistence practices are however on a more fundamental level. This is because the question 'what was the subsistence pattern?' may not be an appropriate one. Essentially, whilst interpretations of Mesolithic subsistence practices tend to be very static, Mesolithic environments would be very variable.

For one thing, environments would vary substantially over different regions. Northern England for example has markedly different topography and geology in different areas, as discussed in chapter two. Most interpretations however tend to ignore regional variability or to simplify it into very basic categories such as upland and lowland, or coastal and inland. Bettinger (1993: 52) provides one explanation for the static and normative nature of many interpretations of subsistence and settlement by explaining that for archaeologists attempting to interpret regional differences 'the easiest way to cope with variability is to ignore it'. Variability in time would also have been substantial. Over short time-scales, differences between 'good' and 'poor' years would affect the relative availability and attractiveness of different resources. Rowley-Conwy and Zvelebil (1989) note the scale of potential short term variability in salmon productivity, and the fact that in certain years certain stretches of river may have been productive, whilst only having low salmon stocks at other times. Other resources may have been equally as variable, especially given the variability now recorded in Holocene climates (Mayewski et al. 1996). The potential effects of this variability on subsistence and settlement system is discussed in chapter four. Over long time-scales however, there would have been marked changes in general climates and in environmental responses. Coastal environments would have been changing fundamentally, with rising sea levels and changing ocean currents, and inland environments would also have experienced fundamental changes throughout the Mesolithic as climates changed and plant and animal species spread gradually from glacial refugia. Although many interpretations of subsistence practices tend to portray a very static picture of Mesolithic economies, some interpretations have incorporated environmental changes into ideas about changing subsistence patterns. These interpretations are however either too broad to apply to a local situation (such as the idea of a diversification of resources discussed in chapter one) or are specific to a single resource (as is Clark's 1972 model of red deer ecology) which may not fulfil the role in subsistence practices which has been predicted. Most fundamentally, most interpretations fail to appreciate that changes in environments through time have different spatial effects, with some regions being more markedly affected by the spread of certain competitive species than others for example. Complex changes in environments can be considered as a major limitation to developing models of subsistence. Alternatively, they may, in contrast, be considered as a core question themselves. It has generally been assumed that, since it is difficult to define precise subsistence patterns at any one time, the effects of changes in environments cannot be addressed. This is not necessarily the case, since changes in environments often cut across a suite of resources. Changes in the character of woodland, for example, would influence the character of many different woodland resources. A decline in nut producing trees or an increase in forest density would affect both small and large game and plant resources. Similarly the gradual in-filling and

eutrophication of inland lakes would, on the other hand, affect a range of different lake resources. Given the limitations outlined, it is clearly not possible to define a precise subsistence pattern at any one time, but it may be possible to discuss in general terms the possible challenges or adaptations that would have been faced, without defining the precise use of different resources (an approach discussed in chapters five and six).

CHAPTER FOUR

Ecological and Ethnographic Analogies

ABSTRACT Our whole understanding of Mesolithic societies has relied, from the very first interpretations of the period, on analogies between modem environments and those in the Mesolithic, or between modem hunter-gatherers and Mesolithic hunter-gatherer groups. Information from both these sources can make a major contribution to our understanding of the period. However, the use of ecological and ethnographic data is often very simplistic. Moreover, misleading assumptions are easily perpetuated where they fit common preconceptions of the period or where the archaeological evidence is not available to refute them (or even appears to be supportive). Many of these assumptions have had a pervasive influence on studies of the Mesolithic, tending to support a very 'static' picture of past activities. In this chapter the origins, development and influence of misleading assumptions, dubbed 'eco-facts' and 'ethno-facts', are considered in tum. It is clear that we understand far less about Mesolithic subsistence and settlement than might first appear from the published literature. More dynamic ecological and ethnographic models are needed, but, since all models are limited by our poor knowledge of past environments, a better understanding of the unique environments in the Mesolithic and how they changed through time is an essential 'first step' in developing a better understanding of Mesolithic settlement (approached in chapter five).

research (as illustrated by Kelly (1995) quoted at the start of this chapter).

ARTEFACTS, ECO-FACTS AND ETHNOFACTS

The development of the two approaches to interpreting the evidence for Mesolithic societies, first the ecological, and then the ethnographic, and their influence on our understanding of Mesolithic settlement and subsistence, are considered in more detail in what follows.

The reconstruction of patterns of subsistence, population and settlement has been one of the key aims of Mesolithic research since the early 1970s. However, it is evident from the distribution and character of the evidence for Mesolithic occupation, as outlined in chapter two, that site-based evidence alone is woefully insufficient to build up a model of subsistence or settlement systems in Northern England. There is little evidence for site seasonality, and the distribution of surviving sites, largely consisting of lithic materials, is much biased by processes occurring at several different scales (as discussed in chapter two). Although site-based evidence is taken into account in approaching past activities, rather than a 'top-down' approach, alternative 'bottom-up' models have usually been based on analogies with present environments, from two different perspectives: •

the resources available in present environments which are similar to those of Mesolithic Europe;

•

the activities of modem hunter-gathers who live in similar environments to those of Mesolithic Europe.

In effect, reconstructions often involve a combination of these two sources. Such analogical approaches depend on similarities between present and Mesolithic environments. They are thus somewhat problematic, since a major limitation that is often overlooked is that Mesolithic environments were unique. Different tree species spread gradually into Britain after the rapid warming at the end of the last ice age, creating somewhat different woodland ecosystems from any we find today (as discussed in chapter three). Modem birch forests, for example, are at present limited to high altitudes and very cold environments, whereas birch forests in the lowlands in the early postglacial flourished in approximately modem-day temperatures. The unique nature of Mesolithic environments also means that commonly used ethnographic analogies, such as the Boreal forest hunter-gatherers of the Canadian Arctic for example, are not necessarily relevant to Mesolithic lifestyles. The unique nature of Mesolithic environments is equally problematic where both direct analogies and general models of environments or of ethnographic societies are concerned. A lack of understanding of Mesolithic environments and the way in which environments changed (particularly how the distribution of different types of environments evolved) is a fundamental limitation to 'bottom up' approaches. A more subtle problem is that, as discussed in chapter one, models are often very simplistic and are rarely subject to critical review. Initial assumptions or suggestions, based on ecological or ethnographic analogies, have often risen to the status of unquestionable 'eco-facts' or 'ethno-facts', which, like Chippendale's (1993) 'factoids' can easily come to dominate interpretations and restrict potential avenues of

52

ECOLOGICAL AND ETHNOGRAPHIC ANALOGIES

THE ECOLOGICAL APPROACH

One of the main characteristics which mark out the study of the Mesolithic from that of other periods is the dominantly ecological approach of much of the research. This approach contrasts markedly with the more sociological or ideological orientation of studies of later periods, particularly the following Neolithic (McCormick and Buckland 1997). Early ecological approaches to subsistence and settlement (such as Clark 1954) were based on quite simple ideas about resource scheduling and seasonal movements of hunter-gatherer groups. However, later developments built on early concepts of resource scheduling by becoming more rigorously mathematical. One arm of the development of ecological models was concerned with specific methods of reconstructing settlement (inspired by patterns of resource exploitation in modern hunter-gatherer groups), and the other explicitly with mathematical methods of reconstructing diets (largely based on models derived from animal ecology). Both of these later approaches were widely influential to broader contemporary interpretations of subsistence and settlement and also to ideas about absolute population and increases in population. It was argued in the last chapter that even simply defining subsistence resources is extremely difficult, however despite these problems quite specific interpretations of settlement have been proposed based on ecological models, and largely accepted uncritically. To understand how and why a number of common conceptions of Mesolithic ecology and society came to be accepted demands a closer consideration of the development and influence of these ecological models.

Clark's Ecological Approach to Settlement and Subsistence After the 1950s, an increasing interest in the development of an ecological approach to prehistoric societies began to develop, with multidisciplinary projects annmg to understand all elements of prehistoric environments, such as the Iraq Jarmo project in the Near East (Braidwood 1974) and the Fenland Project in Britain (Clark 1972; Smith 1997). The excavation of Star Carr (Clark 1954) was at the vanguard of this 'ecological' approach. However, until the late 1960s most interpretations were still dominated by conclusions based largely on the typological study of artefacts. After this time however, academic questions began to be more commonly approached using an ecological perspective, the origins of agriculture being a particular case

Jan

Feb

WINTER

I

Mar

Apr

SPRING

May Jun

in point (Binford 1968; Flannery 1969; MacNeish 1977). Changing approaches to the Mesolithic were very much part of these developments. Both Price (1973) and Clark (1972) stressed that Mesolithic populations were part of an ecological system of dependant parts - an ecosystem - and that a better understanding of this system would make a major contribution to the study of the period. The work of Clark in particular paved the way for an approach that was to become the hallmark of Mesolithic studies. Clark's (1972) model, mentioned in chapters two and three, was one of the first explicitly ecological approaches to Mesolithic resources and settlement, and warrants detailed consideration here. Clark's interpretation of Early Mesolithic subsistence was based on the faunal remains recovered from Star Carr (Frazer and King 1954). Since red deer were the most frequent remains recovered at the site, Clark considered that Mesolithic populations were probably heavily dependant on red deer for food. Clark noted that in Scotland today red deer tend to move between upland and lowland environments from summer to winter. He also noted that known hunter-gatherers tended to move seasonally to follow resources and also that periods of aggregation and dispersal appeared to be a common characteristic of these groups. The contemporary interpretation was that Star Carr was occupied in the winter, and Clark envisaged a clear settlement pattern in which Mesolithic groups would aggregate in the lowlands in Winter and disperse to the uplands in summer following red deer. Clark was even convinced that the same pattern of exploitation would be characteristic of any area with similar topography and that anywhere in Britain 'the recovery of scattered microliths on high ground should prompt a search for a winter base on low ground within the annual range' (1972: 36). Clark's model of Mesolithic settlement has been widely influential. The idea of a winter-summer seasonal movement between lowlands and uplands is a common theme of interpretations of Mesolithic societies (not only in the Early but also in the Late Mesolithic). Jacobi (1978) for example also proposed a model of upland lowland movement in the Early Mesolithic from winter lowland base camps to upland hunting sites following red deer. In this case, Jacobi linked sites at the Lincolnshire edge with those in the Pennines (see chapter 2). Myers (1986; 1989) developed Jacobi's model of upland-lowland contrasts between the Lincolnshire Edge and Pennines, but, in contrast to Jacobi, considered that the upland exploitation of red deer would have occurred in autumn. A similar approach to that of Clark has also been

Jul

Aug Sep

SU.MMER

Oct

Nov Dec

AUTU.MN

WINTER

_______ __ ________

deer huntin~

salmon fishing plant foods __

Figure 4.1

Proposed Late Mesolithic resource availability of north-east Yorkshire (Simmons 1979).

coastal food

used by Simmons (1979), who extended the range of resources considered to include salmon, 'plant foods' and 'coastal foods'. Simmons constructed a simple resource use schedule for the North York Moors in the Late Mesolithic based on a simplification of ethnographic and environmental sources, outlining the supposed availability of the different resource classes according to basic environments. He again predicted that hunter-gatherer populations would have spent the summer in the uplands hunting deer, but with the winter being spent at the coast or in the lowland forest, with periods in spring and autumn on salmon rivers (figure 4.1). Unlike many ecological models, Clark's model has been subject to recent criticism. The concept of a clear dependence on red deer, and also that of upland-lowland seasonal movement have recently been called into question. Legge and Rowley-Conwy (1988; 1989) in particular have challenged the dominance of red deer in subsistence, noting that at Star Carr elk and aurochs provide more meat weight than red deer. They also pointed out that in the forested environment of the Early Mesolithic red deer are unlikely to have carried out upland-lowland migrations. Myers (1986; 1989) in fact also questions the mobility of red deer in his model, suggesting that at least in the Late Mesolithic red deer would have only been present as small dispersed family groups rather than as migratory herds.

Upper edge qf moorland (incl1.1diF19 hazel

Figure 4.2

limitation to this approach is that there is no explicit methodology for defining which resources were important. It was clear from chapter three that the resources considered to be important are often much influenced by the 'spirit of the time' and thus a lack of any means of objectively isolating genuinely important resources presents a clear problem. The simplistic nature of the ethnographic model is also problematic (discussed in the following section on ethnographic analogies). The ecological models which followed Clark attempted to solve the first of these limitations by defining explicitly mathematical means of reconstructing subsistence and settlement. The first of these developments, Jochim's (1976) subsistence-settlement model, has been a direct influence on interpretations of Mesolithic societies. More recent approaches, based on Optimal Foraging Theory, have had a broader influence on interpretations.

More recent models using a similar approach have taken the potential overemphasis on red deer into account. Simmons (1996) uses a range of different resources, including roe deer and aurochs, in his model of subsistence (table 4.1). The basic model of hunter-gatherer mobility, that is to say the idea of longer term seasonal base camps or aggregation sites and short term camps of dispersed members of a larger group has however remained undisputed. Thus, Simmons (1996) used the resource use schedule to suggest a range of 'possible' settlement system models, all of which were based on the long term aggregation site - short term dispersed camps model. The model that he concludes is the 'most likely' model for the settlement-resource schedule for the Late Mesolithic in England and Wales, figure 4.2, consists of a main base camp near the coast occupied for most of the year with a second base camp with subsidiary camps up-river occupied in the summer to autumn period. Season spring summer autumn

winter

Table 4.1

Inland roe deer, pig, plants, eggs red deer, plants, fist aurochs roe deer, pigs, fish, plants, aurochs roe deer, pigs, red deer

Simmons' (1996: 215) 'most likely' model of Mesolithic settlement.

Jochim's Subsistence-Settlement Model and its Descendants From the 1950s onwards, a prevailing optimism about reconstructing subsistence and settlement patterns on the basis of seasonal scheduling was much influenced by research into contemporary hunter-gatherer groups. Some research in particular showed clear quantifiable relationships between environments, resources and hunter-gatherer diets (Birdsell 1953; Baumhoff 1963; Casteel 1972; Keeley 1988; Thomas 1981; Kelly 1995). On this basis, various authors have attempted to apply explicit mathematical models to Mesolithic European data in order to determine the resources which would have been most beneficial to exploit, and ultimately the settlement system.

Coastal eggs, nestlings, fish red deer, freshwater fish, plants fish, plants

seals, shellfish

The first, and undoubtedly the most influential mathematical model of Mesolithic subsistence and settlement was that of Jochim (1976). Since the development of this model, the model itself or various aspects of its construction, have been widely applied to other areas of Mesolithic Europe by authors such as Price (1978; 1980), Tilley (1979), Zvelebil (1981) and Simmons (1996). Jochim studied ethnographic accounts to understand the important factors which determine which resources hunter-gatherers exploit, and

Seasonal resource use model for England and Wales (Simmons 1996: 199).

Clark's approach essentially involved defining settlement patterns on the basis of the distribution of a single key resource, or in the case of later work such as Simmons (1979; 1996) a few key resources. Clearly the most obvious

54

ECOLOGICAL AND ETHNOGRAPHIC ANALOGIES

which factors determine the arrangement of their seasonal exploitation camps and aggregation sites. On the basis of these findings he designed a specific mathematical model which would predict the most likely seasonal resource utilisation and the location of seasonal camps (in effect the settlement system) given adequate knowledge of the characteristics of both the contemporary environment and the available resources (Jochim 1976). The model consisted of three subsystems - the 'Resource Use Schedule', the 'Settlement Location Model' and the 'Demographic Model'. Jochim first demonstrated the utility of the model using an ethnographic example (the Round Lake Ojibwa of eastcentral Canada), and then, using contemporary environments and available resources as the base data, he applied the model to the Mesolithic of south-west Germany.

successfully they fulfil requirements and the distance that people have to travel to exploit them. Jochim assumed that hunter-gatherers would be prepared to travel different distances from a base camp to resources based on the 'pull' of those resources. The relative distance of any settlement from each resource group was calculated on the basis of the dietary proportion of the resource (from the resource use schedule) for each month. The likely location of each seasonal settlement could then be suggested on the basis of the actual geographic location of the primary resource distributions. The 'Demographic Model' was a further step, which assessed the potential for people to aggregate in each season on the basis of the resources available (based on the sum of weight, non-food yields, aggregation and density, divided by mobility for all resources). Settlements in the Resource Red deer Roe deer Boar Beaver Fish Small game Birds Plants

The structure of Jochim' s model was much more complex than that of previous subsistence models applied to the Mesolithic (Clark's 1972 model for example). The first element of Jochim's model, the Resource Use Schedule, was based on defining the important 'goals' (on the basis of ethnographic accounts) which governed the decisions huntergatherers made about which resources to exploit. Jochim identified two main goals - the attainment of a secure level of food and manufacturing needs, and the maintenance of energy expenditure within a predefined range, determined partly by the need for population aggregation (Jochim 1976: 25-26). The taste of different resources (largely governed by the fat content), the variety of resources exploited and the prestige associated with exploiting a resource were factors that were also important, although Jochim chose to consider these factors after the resource use schedule was constructed. The important characteristics of a resource in relation to these two goals were identified as the weight, density, aggregation size, mobility, fat content and non-food yields. Resources would 'score' highest for the first goal - secure income of food and manufacturing items - with greatest weight and non-food yields, with greatest density (as risk of not capturing or collecting the resource decreases) and with the lowest mobility (where risk of not capturing or collecting is again the least). With regard to the second goal, a resource would be less expensive in terms of energy expenditure the greater its yield and aggregation size and the lower its mobility. The relative benefits of each resource at any month (for the two combined goals) could thus be calculated to give a measure of the importance of each different resource type in that month. By totalling the contribution of a resource to the diet for each month, and then calculating the fraction of the cumulative 'attraction', a model of the 'predicted distribution of utilisation' - the seasonal exploitation of each resource - could be generated.

Table 4.2

Percentage of yearly diet 26 3

22 1 13 13 2

20 Percentage contribution to diet of major resource groups (Jochim 1976: 108).

season with the highest potential for aggregation would be expected to be aggregation sites. In order to apply the model to south-west Germany, Jochim collected data on the ecology of different resources to generate figures for weight, density, aggregation, mobility and non-food yields for each resource. In effect, detailed figures (described in chapter three) were only available for large game where each species could be considered separately. Other groups of resources - fish, small game, plants, birds - were considered in very general terms. In fact, both plants and birds were given a 'dummy' figure rather than being calculated on the basis of the attributes, and small game 'filled in' where other resources were limited. Jochim' s predictions of the overall contribution of resources to the diet in Mesolithic south-west Germany are shown in table 4.2. A graph of the year round 'predicted distribution of utilisation' of resources was then generated (figure 4.3).

Jochim suggested that this resource schedule, and the 'pull' of various resources in the 'second stage' of the model, could be used to predict the patterns of the settlement system. This could be done via a 'Settlement Location' model and a 'Demographic' model. The Settlement Location Model was based on 'gravity' models of spatial attraction used extensively in geography to predict patterns such as the movements of consumers to new shopping centres (for example Foot 1981; Haynes and Fotheringham 1984; Birkin and Clarke 1991). The principle of these models is that resources are more or less attractive depending on how 55

(1976) to the Mesolithic of the Netherlands and Tilley (1979) constructed a similar model for the Mesolithic Communities of the Fenland edge of East Anglia. These later models have used a similar approach with progressively less rigour. It is noticeable that Price (1978; 1980) used far less detailed ecological studies than those of Jochim with little attempt to assess the archaeological or ethnographic support for the model. Nearly all elements were remarkably similar to Jochim's model (aurochs, red deer, wild boar and small game are calculated as each contributing about 15% of the diet) although Price 'guessed' bird contribution at 5% and plants at 15%. Unsurprisingly, the final model of seasonal resource exploitation was remarkably similar to Jochim's. Tilley' s model was based on insight rather than mathematical comparisons, he commented that 'A nonquantitative use of this general model is applied here' (Tilley 1979: 15). Thus when Tilley considered the types of available resources and estimated their percentage contribution to diet, he proposed a higher contribution of plant foods to the diet (in the wake of Clarke's 1976 paper), and rather lower contribution of both boar and fish. Inevitably, the differences between Tilley's model and Jochim's model are more related to individual opinions about which resources ought to be important than genuinely reflecting different environments.

.40

,g E

~

\,

.30

' ,_

-0

ci

c:;

.2 e:; .20 0

Small

C4 0 '-, A..

.10

F

Figure 4.3

M

A

M

J

A

S

O

N

D

Model of seasonal resource exploitation in the Mesolithic of south-west Germany (Jochim 1976: 115).

Jochim simplified the predicted seasonal exploitation of resources into a series of six 'seasons' of resource exploitation for south-west Germany (table 4.3). He supported the seasonal distinctions proposed by reference to the exploitation patterns of the Salishan Indians of the North American interior, living in an environment including mixed forests, dry grassland and subalpine communities (Jochim 1976: 116).

Other later approaches have again drawn on Jochim's model although less explicitly. Zvelebil (1981) compared the site catchments of various sites in prehistoric Finland through a similar resource use schedule, but used this as a basis for catchment analysis (based on the productivity of different resources). In this case the resource use value (calculated bimonthly) was constructed from measures of reliability of capturing prey, non-food and nutritional yields, procurement and transport costs, as well as yield and storage potential (Zvelebil 1981: 183). Once again, large game were the only resource for which detailed data was available upon which to make judgements of relative importance.

According to Jochim' s Settlement Location model and Demographic model, base camps in Mesolithic south-west Germany would have been located close to each other on the floor of the main Danube valley, with either a two base camp or four base camp system. This observation fitted well with ethnographic evidence of hunting groups typically travelling farther to the best areas for hunting large game, with base camps being located nearer more reliable and bulkier resources, a model which Binford (1980) came to define as Logistical Foraging.

After Jochim's model, alternative approaches to modelling past subsistence practices were developed. In the 1980s a new series of models came to the fore, these new approaches offered the potential for assessing the relative importance of different groups of resources both objectively, and more simply. These models, derived from the discipline of animal ecology, are subsumed under the title of 'Optimal Foraging Theory'.

Jochim' s model is often taken as the 'final word' on Mesolithic subsistence practices, not only for south-west Germany but also for elsewhere in Europe. Although Jochim commented that 'the patterns of subsistence and settlement formulated for the Late Mesolithic of Germany are adapted to this region and time period and should not be generalised'(1976: 187), the resource use schedules used in later studies have been remarkably similar to Jochim's model. Price (1978; 1980) applied the same model as Jochim

JanFebMar red deer boar small game

Table 4.3

April plants deer fish

May fish plants deer

JunJulAug plants fish deer boar

Resource exploitation seasons (Jochim 1976: 116).

56

SeptOct plants deer boar fish

NovDec boar deer plants

ECOLOGICAL AND ETHNOGRAPHIC ANALOGIES

are those not aiming to reconstruct completely the range of resources used, but instead addressing the likely effect of changes in environment or exploitation (such as changes in technology). Hill and Hawkes (1983) for example showed that, as expected, the Ache decreased their diet breadth when shotguns were adopted to hunt animals, dropping lower return resources such as monkeys and small birds out of their diet. Early applications of optimal foraging theory used energy as the 'currency' comparing the relative energetic costs and benefits of acquiring different resources. However the successful survival of hunter-gatherer populations is often dependant on other criteria (discussed in chapter three). For example, other characteristics than calorific yields may determine which resources are preferred. Speth and Spielman (1983) stress the importance of fat in huntergatherer diets at times when only very lean meat is available (the protein metabolism problem), thus hunter-gatherer populations are known to frequently crave fat when supplies are short (Kelly 1995). Not only fat but also protein, vitamins and minerals are as vital to diets as energy. Bailey (1975) points out that the daily requirements of almost all nutritive substances can be obtained from only a small amount of shellfish, counterbalancing their low energetic returns.

Optimal Foraging Theory 'Optimal Foraging Theory' gained major recognition through the work of Stephens and Krebbs (1986). Though the ideas were widely influential, Winterhalder and Smith (1981) played an important role in exposing archaeologists to these models, and Mithen (1987; 1989; 1990) to applying aspects of the theory to the Mesolithic. The central assumption of optimal foraging theory was that huntergatherers choose from available resource options in order to optimise certain criteria (often maximising calorific yield compared to calorific expenditure) whilst being subject to certain constraints. Whilst Jochim' s model considered the relative benefits of exploiting different resources in terms of their abundance and costs in terms of 'search time' (through the density and aggregation attribute), models from optimal foraging theory take into account the absolute 'cost' of both procuring and processing a resource (for example, costs of collection and processing can be considerable for some plant foods). The Viet Breadth' model, used to determine the relative importance of different subsistence resources, has been the most important influence on discussions of Mesolithic subsistence. Other components of optimal foraging theory, such as the 'Patch Choice' model, defining the length of time a 'patch' of resources is exploited, have not been explicitly applied to the Mesolithic situation.

Furthermore, certain resources may be important because, even if their calorific or other yields are low, they are reliable, as noted in chapter three. The exploitation of plant foods incorporates a much lower risk of coming home 'empty handed' than hunting (particularly when hunting large game). On the other hand, by taking risks, huntergatherers may reap higher benefits overall. Mithen (1990) explores the issue of risk and the role of prehistoric decisionmaking. He suggests that a diverse range of resources at coastal sites allowed Mesolithic hunters in these locations to gear their hunting strategy to exploiting large game more efficiently (they could afford to take the risk of failure as other resources were always available), whereas in inland areas (in this case south-west Germany), they would have to avoid returning empty-handed since they had fewer 'back-up' resources (such as marine fish or plant resources) (1990: 89193). He compares Upper Palaeolithic hunting strategies at the Spanish site of La Riera with Mesolithic strategies at coastal sites in Denmark, concluding that at the latter, hunters were indeed able to concentrate on more 'risky' resources because 'back-up' resources were available if they came back empty-handed (Mithen 1987; 1990).

The Diet Breadth Model The most relevant element of foraging theory to discussions of past subsistence resources is the 'Diet Breadth Model'. This model assumes that hunter-gatherers will choose which resources to exploit on the basis of their 'return rates' (the overall benefit of exploiting the resources taking into account the time taken to exploit and process them). Resources are ranked in terms of their return rates (post encounter), search, pursuit, capture, transport, processing and handling time. The highest ranking resources can be analysed on a seasonal basis to determine the 'optimum' seasonal resource exploitation. Although hunter-gatherers do not make explicit decisions in this way, research has shown that subsistence and settlement behaviour can be explainable in terms of such optimal use of available resources. O'Connell and Hawkes (1981; 1984) and Hill and Hawkes (1983) found some success with using the diet breadth model to predict the resources which the Alywara (in Central Australia), and Ache (in Eastern Paraguay), exploit.

The most important conclusion of the diet breadth model is that a resource's abundance alone cannot be used to predict whether it will be utilised. Resources which have low return rates may be ignored if other resources which provide higher 'returns' on investment are available and fulfil requirements (acorns may have been abundant in oak forests in the Mesolithic but this does not necessarily imply that they were an important food source as they are time consuming to collect and process). Rowley-Conwy (1986: 27) illustrates this point, noting that highly productive resources such as earth-worms are not exploited because hunter-gatherers clearly take search and processing costs into account rather than exploiting resources solely on the basis of their abundance. The diet breadth model also predicts that an increase in the search costs of high ranked resources (such as large game) may cause lower ranked resources to be included in the diet. Perhaps the most interesting applications

Rather than any specific model, foraging theory has perhaps been more influential in forwarding the understanding of the range of factors which might influence any subsistence or settlement pattern. The range of criteria which influence decisions about which resources to exploit are clearly very varied (and may change radically over time). In certain cases it is even non-food resources which may determine settlement patterns, and thus the food resources exploited are determined by the location of settlement rather than vice versa. The availability of water or firewood may be more important than the energy efficiency of foraging in certain conditions, especially in cold or arid regions (Kelly 1995). Although attempts have been made to combine different criteria to address the ranking of resources (Hill 1988), this 57

Jan

Feb

Mar

Apr

May Jun

Aug Sep

Jul

Oct

Nov

cod ___________

-----------mackerel

eels small whale harp seal

________ ___

grey seal pups

_____ ________

Figure 4.4

_____

_____________ ________ ________ oysters

fur animals

land mammals _____ swans ~-~--ducks hazelnuts acorns

_____ _____

________

Dec

fruits plants cockles mussels

Proposed resource availability schedule for the Danish Erteb0lle (Rowley-Conwy 1984).

Jan

Feb

Mar

Apr

May Jun

Jul

Aug Sep

Oct

Nov

Dec

--------------------------------

shellfish ___________

_______ _____

plant foods

adult sea birds sea birds eggs _____________

salmon

----------------________ __________

waterfowl -----

Figure 4.5

estuary /inshore freshwater fish offshore fisl

land game

Proposed Late Mesolithic resource availability in the Eskmeals area (Bonsall 1981: 466).

type of approach is undoubtedly extremely difficult given the range of important characteristics. In fact, perhaps the most significant development to stem from foraging theory has been the growing recognition that the factors determining prehistoric diets would have been extremely complex.

Jochim's (1976) model, perhaps because the conclusions of the latter, presented as a clear graph, have been easier to apply as a piecemeal 'eco-fact'.

Of the interpretations that have been influenced by developments in foraging theory, possibly the most notable are discussions of the subsistence patterns of potentially sedentary coastal populations. At the coast, as discussed in chapter three, abundant migratory resources are potentially available at little cost. The year round availability of resources with high return rates has been a key element in the argument for sedentary complex coastal communities. Rowley-Conwy (1984) for example used return rates in considering which resources were available at coastal sites in the Danish Ertebjljlle (although not quantitatively) with largely only the 'high-ranking' resources included in the resultant resource-availability schedule (figure 4.4). Bonsall (1981: 466), figure 4.5, also represented resource availability in this way for Late Mesolithic populations at Eskmeals in Cumbria. In general terms however, the influence of foraging theory, a whole body of theory based on developments in ecology, is much less visible than that of

58

ECOLOGICAL AND ETHNOGRAPHIC ANALOGIES

THE LIMITATIONS TO CURRENT ECOLOGICAL MODELS OF SUBSISTENCE AND SETTLEMENT

The Source Data The influence of the above models has been far-reaching. Explicit models are often taken as the 'last word' on subsistence, and even where this is not the case, their components, a heavy reliance on large game for example, are frequently a common preconception brought to bear on understanding the period. In fact, each successive model of subsistence and settlement, whilst also incorporating el~ments of contemporary preconceptions, has clearly been bmlt upon the conclusions of previous models. Jochim' s proposals about the relative contribution of different resources thus present us with prime examples of 'eco-facts' since,. without any serious criticism, they were used directly Pnce (1978; 1980) and Tilley (1979) and also clearly mfluenced many later interpretations including Zvelebil (1981) and Simmons (1979; 1996).

?Y

The most obvious limitations to models of subsistence and settlement which are built on the basis of resource availability are, as outlined in chapter three, the limitations to our knowledge of past environments, and of how Mesolithic populations exploited them. Mesolithic environments were not only unique, with no modem parallels, but the evidence for the presence and availability of different resources (particularly resources other than large game) in these environments is also very limited. Equally, even if we had a perfect knowledge of past environments, Mesolithic subsistence would still not be predictable. Populations could have e~ploited their environments in many different ways, dependmg on the use of specific technologies (such as fish nets or traps for game), on how intensively resources were exploited (determined in part by the population pressure on resources as well as other factors) and whether characteristics such as storability or fat content were important (as discussed in chapter three). Even quite simple interpretations of the most important res~urce groups in fact depend on a detailed knowledge of e~vironments and exploitation patterns. Like Clark, Simmons (1979; 1996), constructed his settlement models simply on the basis of defining a few key resources and their s~asonal availability and distribution. He included not only different types of large game at different seasons (red deer, roe deer and aurochs) but also in the later model includes other resources such as eggs and nestlings as key resources ~Simmon~ 1996). This model may be attractively simple, but ~s w~rrymg dependant on the correct resources being identified. The resources which appear to be most obvious to ourselves can be very different from those exploited by past or present hunter-gatherers. Jochim's (1976) model may have been less subjective than Cl~k'_s (1972) approach, but it is still far from a clearly objective approach. Though widely accepted, even Jochim's (1976) model is much biased by a series of assumptions about resources and by problems with a lack of evidence for past resources or exploitation patterns. As Zvelebil (1994: 58) points out, studies of the relative benefits of plant and animal exploitation are not available for temperate

woodlands akin to those in Mesolithic Europe. Thus, Jochim guesses the relative contribution of plant foods to diets as 20%. The contribution of birds is also guessed at 2%. Fish resources were compared quantitatively with other resources but with difficulty, since measures such as density and mobility are difficult to define for fish (Jochim rates fish ~ability as extremely low as they cannot escape from the nver). The factor of fish resources is also made more problematic since different technologies (nets or weirs) would have had a major impact on the relative benefits of fish exploitation. Small game resources were perhaps the most problematic being even intriguingly accorded a 'fill in' status - their use being inversely related to the utilisation of other classes of resources. The least subjective element of Jochim's (1976) model was large game resources, for which detailed information on densities and food and non-food values in modem forest ecosystems was available. It is probably only because the comparisons between different large game components were the most reliable aspect of this model that the model was successful at predicting Ojibwa subsistence practices. Even for large game resources however the figures used by Jochim are _disputable. Bay-Peterson (1978: 123) for example publishes figures for ungulate densities that vary by as much as a factor of ten. In fact, because of the problematic nature of_the evidence for present large game availability in Europe, Mithen (1990) had to use data on ungulate availability in East Africa in his foraging theory based model of Mesolithic hunting tactics. Even given the best information and the most careful and detailed analysis, the issue of defining resources can be more complicated than it first seems. The abundance of any resource can vary widely across quite limited scales of time and space, moreover modem ecological work suggests that animal populations can be stable at different levels in the same environments depending on the past history of predator-prey relationships (Flowerdew 1987: 130-135) a phenomenon known as the 'multiple stable state'. However, the problematic nature of the evidence for resource availability and use used in Jochim' s model, and thus in his conclusions, was rarely considered by later similar models (such as Price 1978; 1980; Tilley 1979 and Zvelebil 1981). Fortunately, the seasonality (or seasonal availability) of different resources is somewhat less prone to problems with non-analogous environments than their abundance - plants flower and seed, animals breed and hibernate, and birds and sea mammals migrate at what can be assumed to be broadly analogous times in the present and the past. Resource availability schedules can thus provide a useful tool for studying the seasonal scheduling of past hunter-gatherer groups. There are a number of limitations however. Resource availability schedules cannot be used to determine if resources were important, as they give no indication of the abundance of resources (only their potential availability). Bonsall (1981: 466), figure 4.5, for example, shows salmon at Eskmeals being available for five months, whereas they wo~ld only be significantly abundant for a very limited penod - the salmon 'run' - within this time. Similarly, resource availability schedules can also differ depending on source data. Cod and mackerel are available from May to

August according to Rowley-Conwy (1984), figure 4.4, but the exact opposite (as 'offshore fish') of September to February for Bonsall (1981), figure 4.5. The year round availability of resources is often used specifically to suggest that since sufficient resources were available local populations may have been sedentary. There are a numbers of problems with these inferences. First, resources may not abundant or productive enough to support populations - for example Rowley-Conwy's (1984) suggestion that oysters may have been a vital poor season resource for the Ertebjljlle is debatable, given their low calorific content and high processing costs. Secondly, even abundant 'high-return' resources which are available all year round are not necessarily linked to sedentary or complex societies - Schalk (1981) demonstrates a wide variety of settlement behaviour despite very similar availability of coastal resources along the north-west coast of North America. Equally, many societies are sedentary where resources are not available all year round by virtue of systematic storage of resources available at other times.

Spatial and Temporal Variability 'Bottom up' models suffer not only from limitations with evidence for the availability and use of different resources, but also from problems with the way in which ecological models are constructed. The most serious limitation is that most models present an almost exclusively 'static' model of past subsistence and settlement. Spatial and temporal variations in resources can be significant on both small and large scales. Early Holocene environments were not only very variable across different landscapes - uplands, lowlands, coasts, rivers - but also changed markedly through time as different tree species spread from glacial refugia, and climatic fluctuations took place. Thus a substantial variability in subsistence and settlement behaviour ought to be expected. In contrast, one of the main aims of ecological studies has been to define the subsistence or settlement system. As a result of the static emphasis of most, if not all, ecological models, some of the most interesting aspects of Mesolithic subsistence and settlement - the potential responses to marked changes in early Holocene environments - are often overlooked. Some of the most fundamental limitations of ecological models, in particular the static or normative nature of most models, have been much influenced by the application of concepts derived from ethnographic sources. These sources are largely much oversimplified and moreover place a constraint on possible interpretations. But, before considering the use, and misuse, of ethnographic analogies in the following section, the use of ecological determinants to reconstruct population numbers is considered below.

POPULATION Only a limited number of authors have attempted to define absolute population numbers on the basis of resources; most authors prefer to draw on ethnographic rather than ecological parallels to suggest population numbers. Where specific population numbers have been suggested, the idea of a large game hunting base to subsistence practices (discussed in chapter three) has once again been very influential. Clark (1972: 38) for example, based his figures for population density on the yield and meat weight of deer (from studies in the Scottish Highlands) and human calorific requirements. He concluded that there would have been 3,300-8,800 people in England and Wales in the Mesolithic, that is a density of 0.03-0.07 persons per km 2 • More recently, Smith (1990: 14) also assumed a dominant role for meat from large game animals in the diet, although he somewhat more cautiously used large game densities to determine maximum population densities, that is the 'carrying capacity' of the environment. He used measures of ungulate biomass and potential yields in present boreal and temperate deciduous forests, calorific yields of ungulates and calorific requirements of human groups to calculate population densities. Smith concluded that there would have been a maximum population density in the Mesolithic of 0.05 persons / km 2 for boreal forest and 0.16 persons/ km 2 for temperate deciduous forest (with an unstated implication that populations would increase as woodland types changed through time). Aside from the problems of using large game densities from modem woodland environments (or the Scottish Highlands) in the above cases, a more fundamental problem is that studies of the ecological determinants of historic huntergatherer populations suggest that it is the bottleneck of the resources at the poorest season (rather than overall yields of any resource) which determine population numbers (as discussed in chapter three). This relationship was also noted by Jochim (1976: 134) and Mellars (1975: 54). Thus measures of the carrying capacity (such as those of Clark and Smith above) which do not take the seasonality of the environment into consideration will greatly overestimate the real population in an ethnographic situation. Casteel (1972: 27-35) showed that estimates of carrying capacity on the basis of year-round resource yields were 20-25 times the actual ethnographic figures for many New World groups. He demonstrated that for the Chipewyan, fish, the main 'poor season' resource, could be used to calculate maximum population. Baumhoffs (1963) study of historical Californian populations also showed that in the Lower Klamath province, fish yields were the best predictor of population numbers. Jochim proposed that fish (as the 'lean season' resource in his subsistence model) would be the main determinant of population numbers in Mesolithic south-west Germany. He suggested that the upper limit for population would be 0.13 persons per km 2 (Jochim 1976: 135) but that, given factors such as the structure of the river basin and the organisation of hunter-gatherer groups, the actual numbers might be much less. Jochim' s approach is more clearly related to ethnographic studies, however there is no a priori reason for fish to have been the key poor season resource in Mesolithic south-west Germany. Clearly the importance of fish resources varies between different

ECOLOGICAL AND ETHNOGRAPHIC ANALOGIES

hunter-gatherer groups, and in other regions of California Baumhoff (1963) demonstrated that a combination of three resource categories - fish, acorns and game - could be used to predict populations, depending on the relative role of these resources for hunter-gatherer groups. As noted in chapter three, defining the relative roles of different resources is extremely difficult.

explanation - and the goal of the latter applications of ecological models to the Mesolithic situation - prediction are very different things. Given the problems outlined in chapter three and above, the optimism for predicting of the past subsistence and settlement patterns on the basis of environments may have been misplaced. This is not to imply that the understanding on past environments is not essential to any understanding of Mesolithic societies, if not a means of reconstructing the settlement pattern, it is a crucial component of any attempt to understand variability or change in past societies.

It is notable that very few authors deal with a change in absolute population numbers, although many suggest that populations were gradually increasing (e.g. Newell 1973; Meiklejohn 1978: 75; Morrison 1980: 136; Price 1983; Gendel 1984; Vang Petersen 1984; Myers 1986; 1989: 89; Verhardt 1990; Smith 1992). Ecological changes are normally seen as the prime motivator for changes in population (as discussed in chapter one) and several authors interpret changes in woodland types in the Mesolithic (from boreal to temperate woodlands) and warming climates in terms of an increase in populations both in Britain and more widely in the rest of Western Europe (for example Jacobi 1978; Myers 1989; Rowley-Conwy 1983). In a later article partly based on his 1976 volume Jochim (1989), in contrast, suggests that populations during the Mesolithic in south-west Germany would actually decrease as forests increased in density, and undergrowth plants and forage for game animals was reduced.

One of the key lines of support for the concept of the definable subsistence and settlement system have been ethnographic analogies. The development of influence of models of Mesolithic subsistence, settlement and population based on ethnographically documented hunter-gatherers is considered in the following section.

The range of interpretations both of absolute populations and of changes in population illustrates that measures of population numbers are very problematic, and measures of changes in population more so. Certain critical 'poor season' resources may have determined Mesolithic populations but, as noted in chapter three, it is not clear which resources these might be, and even if these resources were definable, the yields are similarly equivocal. Successful determinations of population numbers have only been made where both the available resources, and methods of exploitation, are known in detail (i.e. in cases of modern populations where detailed ethnographic records exist and modern environments are a suitable analogy for near modern resource availability). Without the knowledge that acorns were intensively processed, for example, estimates both of critical resources and of resource yields for Californian populations would be very different and unlikely to be as good a predictor of aboriginal populations as Baumhoff (1963) found. Clearly it is difficult to separate the influence of ethnographic concepts from ecological ones, whether of subsistence and settlement or population. At a more fundamental level, it was the ethnographic analyses of the relationships between environment and subsistence and settlement in known hunter-gatherer groups (Birdsell 1953; Baumhoff 1963; Casteel 1972; Thomas 1981; Keeley 1988; Kelly 1995) which first provoked a pervasive optimism about reconstructing subsistence, population and settlement on the basis of past environments. The formulation and acceptance of many models applied to the Mesolithic followed ethnographic analyses which demonstrated clear relationships between environments and subsistence and settlement. That either subsistence or settlement patterns should be clearly predictable, given a detailed knowledge of environments, is itself an eco-fact. An essential problem with this inference being that the goal of the former analyses 61

THE ETHNOGRAPHIC APPROACH

Figure 4.6

The Selk'nam of Tierra del Fuego: Hunteraatherers in a forested environment.

Almost from their first discovery by Europeans, modem hunter-gatherers on distant continents have been seen as a source of evidence for the lifestyles of past populations (Orme 1981; Schrire 1984). There are many early accounts of hunter-gatherers as examples of our ancestors 'frozen in time' range (from Lubbock 1865; Pownall 1795; Wilson 1851 to Clark 1951 - quoted at the start of this chapter). Although most modem ethnographic studies no longer see recent hunter-gatherers as a direct parallel for past societies, ethnographic research still has a major impact on archaeological interpretations. Ethnographically recorded societies (such as the Selk'nam shown above, figure 4.6) continue to be used as an analogy for Palaeolithic and Mesolithic populations, either directly as 'piecemeal analogies' (Orme 1981), or through approaches such as general observations, applied models and extrapolations from statistics. Modem hunter-gatherers provide us with a valuable record of a range of subsistence and settlement patterns of mobile foragers, and a means of structuring distinctions evident in archaeological evidence, but ethnographic analogies have to be used with care. The surviving record of hunter-gatherer groups is a very biased one, not only towards North America, but especially towards the most remote or inaccessible locations on that continent. In effect the areas with particularly extreme environments, especially those marginal for agricultural exploitation in the farthest west and north, and those where economic activities (such as trapping) allowed a later survival of hunting and gathering, were the last to be affected by European colonists. Those were thus the areas where most research has been possible. Kelly's (1995) survey of available data on hunter-gatherers, for example, drew on data from 129 societies, but of these 60 were from North America and 31 (a quarter of the total) from societies on the west coast of this region (the area which was colonised latest). Additionally, because of the history of colonial contact, assessments of population density (in the wake of diseases such as smallpox) and settlement patterns are thus very biased. Post-contact populations, in effect, are neither representative of past variations, nor can they be

considered 'pristine' and uninfluenced by W estem contact. Arguments thus arise over which aspects of activities have a long history and which relate to colonial impact. The short timespan of research also tends to present a very static picture of these societies (Jochim 1988) and research concentrated in specific regions tends to underestimate long distance influences and exchange (Wobst 1978). In fact, Wobst even refers to the problems of a dependence on biased ethnographic accounts emotively as the Tyranny of the ethnographic record in archaeology'. Unfortunately, the above biases are but rarely considered in the use of ethnographic analogies for Mesolithic societies. Not only that, but also the relevance of ethnographic analogies, (for example the similarity of present environments to those of the past) and the scope (or specificity) of ethnographic generalisations are often given little attention. Although a knowledge of ethnographically documented societies can make made a major contribution to our understanding of the period, it is possible to trace a growing acceptance of initially carefully voiced, but in many cases misleading, assumptions which are often based on simple analogies. These assumptions, which have grown to constrain our understanding of the Mesolithic, are here dubbed 'ethno-facts'. The development of approaches drawing on ethnographic analogies and the rise of such 'ethno-facts' are considered below for models of population and for models of settlement systems. Whereas population estimates are the 'last phase' of ecological models, they are often a starting point for models of subsistence and settlement based on ethnographic analogies. As such they are considered first below.

ECOLOGICAL AND ETHNOGRAPHIC ANALOGIES

POPULATION As illustrated in chapter one, population has been seen for some time as a major structuring element in social systems, and population change as a prime mover for social change. Meiklejohn (1978: 68) comments that 'Of the variables that most control the social systems of any group, population size is the most important'. Newell and Constandse-Westermann also note that 'Population density is directly related to the level of social organisation' (1986a: 276), and further relate population density to technological complexity (1986b). Observations of the population density of known huntergatherer groups are often used as a basis for suggested population densities in the Mesolithic, and also serve to support models of Mesolithic populations derived from other sources (typically resources or site-based evidence). Clark (1972: 28) was probably the first to make direct comparisons between ethnographic population densities and those in the Mesolithic when he used population densities of people in Ta~mania (from the Chambers encyclopaedia, 1950: 473b) to estimate total population numbers in England in the Mesolithic (his estimates were between 4,133 and 10,455 people). He claimed that this was a 'reasonably good' (Clark 1972: 28) match with his figures derived from a dependency on deer (which ranged from 3,300 to 8,800 people). Somewhat later, Meiklejohn (1978) also supported his population estimates by referring to ethnographic figures. His suggested that his estimates of Upper Palaeolithic populations in Northwest Europe (between 0 008 to 0.090/km 2) were reasonable as they were 'well within the range of densities known from modem hunter-gatherer populations' (Meiklejohn 1978: 70). These figures were however derived from what we would now consider to be a very suspect source - the numbers of archaeological sites (after Bordes 1968). Meiklejohn further used a population growth rate of 0.004-0.01 % per annum to calculate Mesolithic population based on these estimates - giving overall increases of between 27% and 332% (and population densities thus of 0.0l-0.3/km 2). 'Ethnographically derived' figures for Mesolithic populations appear in a number of publications. However, in most cases apparently ethnographically derived figures are based on secondary sources (and Meiklejohn's (1978) Palaeolithic estimates have been particularly influential) with these figures being 'passed down' from publication to publication. Newell and Constandse-Westermann (1986a), for example, used Meiklejohn's Upper Palaeolithic population density figures, alongside unspecified growth rates, to suggest gradual increases in population throughout the Mesolithic. Smith used an 'ethnographically derived' figure of 0.012 people per km 2 (Smith 1990: 16) to support his estimates of population density on the basis of available resources, and Simmons (1996) states that from 'analogies with near recent populations of hunter-gatherers' the range of population density in the Mesolithic would be 0.01-0.1 people per km 2 (Simmons 1996: 161). The problem with using any records from ethnographically documented societies to estimate Mesolithic population

numbers is that the range of variation in known huntergatherer population densities is very large - roughly a thousand-fold difference in densities is recorded. Thus, almost any figure can be within the range recorded, and moreover some society may be found to support almost any population estimate. Recorded densities are in any case problematic, with records of densities largely taken after warfare and diseases have affected indigenous population numbers. Also the issue is further complicated since we have no records of hunter-gathers in analogous environments today to use as an analogy for Mesolithic populations. It would be tempting to believe that Mesolithic populations were really constrained by the densities defined. However, where surveys of hunter-gatherer densities have taken place, they illustrate the wide range of variability expressed, rather than the predictability. Thus the population densities documented in Newell and Constandse-Westermann's (1986a) analysis of population densities of 169 North American ~unter-gatherer societies range between 0.002 and 63.096/km . More recently, Kelly (1995: 221) illustrated that for hunter-gatherer societies for which information is available (205), population densities are extremely variable (only 69 (34%) lie within the 0.0l-0.l/km 2 range). Clearly figures for absolute population densities and for changes in population have been substantiated through misleading ethnographic analogies. The idea that Mesolithic population densities can be taken to lie neatly within the 0.0l-0.l/km 2 range is thus unfortunately an example of the development of an 'ethno-fact'. That is, an initial assumption which is taken as a 'given' by later authors and is replicated 'down the line' of later publications, in much the same way that Jochim' s estimates of resource contributions took on the status of 'eco-fact' in the absence of contradictory (or supporting) evidence. Unfortunately, although equally (and perhaps more obviously) biased, models based on 'ethnofacts' have become more popular in recent years. This is at least in part because ecological 'resource-up' models have fallen out of favour, partly because of a lack of evidence for resource availability and exploitation patterns, but more particularly because mathematical and 'deterministic' approaches have become increasingly criticised. The development of these ethnographic models of settlement and the further rise of the 'ethno-fact' are discussed below.

467-72). He also noted that 'subsistence units' were part of a wider population which would gather together at some time in the seasonal cycle for as long as resources allowed.

SETTLEMENT SYSTEMS

A number of different 'ethno-facts' have had a major influence on reconstructions of subsistence and settlement patterns in the Mesolithic. Nonetheless, the origins and development of the 'ethno-facts' themselves is difficult to isolate. This is not least because from the earliest interpretations of prehistory, accounts of the lifestyles of living hunter-gatherers have coloured and structured interpretations of the past (Orme 1981). In some cases, with the use of direct analogy, the role of interpretations of ethnographic societies in structuring understandings of the Mesolithic are clear. However, ethnographically documented societies structure most interpretations of Mesolithic rather more subtly, through the use of a general model of huntergatherer settlement, in some cases explicitly defined, while in others being more of a 'taken for granted' preconception of hunter-gatherer existence.

Having assumed that boreal forest environments and resources were similar to those in Mesolithic Europe, Price (1973) also assumed that the means of exploitation of these resources would also have been similar. He remarked that 'Mesolithic populations must have been dispersed over the landscape in relatively small subsistence units' and 'agglomeration should be expected to occur sometime in the yearly cycle when sufficient resources are available ... A cycle of seasonal activity is predicted' (Price 1973: 472). Price suggested that the archaeological record be studied for information on group size and activity to illustrate these patterns. Price's (1973) interpretations were clearly somewhat problematic, not only because Mesolithic environments were very different from those of the Canadian sub-Arctic (as discussed in chapter three and five) but also because such 'piecemeal analogy' (Orme 1981) is easily biased by the particular historical or social context of the group being studied. Nonetheless, Price did make some important points, particularly by drawing attention to the complexity and variability of ethnographically recorded settlement patterns. He noted that the Cree and Ojibwa used a series of different types of residential camps and short and long term occupations to exploit different resources during the suggested annual round for example. He also commented on changes in resource availability (in this case fluctuating rabbit populations) which affected settlement patterns, and the fact that procurement and settlement systems varied even between groups in apparently similar environments. The complexity of known hunter-gatherer settlement systems noted by Price is, in contrast, often overlooked in later studies.

Direct Analogies One of the first explicit uses of ethnographic analogy to interpret evidence for Mesolithic economy and society was put forward by Price (1973). Price made a direct analogy between Canadian sub-Arctic groups, living in a boreal forest environment, and hunter-gatherers in Mesolithic Europe. He drew on evidence from three hunter-gatherer groups - the Mistassini Cree (Rogers 1963; Rogers and Rogers 1959), the Attawapiskat Cree (Honingmann 1956; 1961) and the Round Lake Ojibwa (Rogers 1962) whose 'procurement system' might serve as a model for the Mesolithic of Europe. Price outlined the relative contributions of the main groups of resources to their subsistence base as recorded by Rogers (1966; table 4.4). On the basis of the diets of these groups he suggested that 'there is no evidence to suggest that the collection of plant foods provided a significant proportion of the diet' in Mesolithic Europe (1973: 472). However in contrast to Clark's (1972) predominant emphasis on large game, through analogy with boreal forest groups, he suggested, rather than purely a large game emphasis that 'Hunting of large and small game, fishing and some fowling seem to have provided the subsistence base' (1973: 472).

A General Model of Hunter-Gatherer Settlement More than direct analogies, the application of general concepts of hunter-gatherer settlement are one of the most common uses of ethnographic evidence. These models are supposedly drawn from a compilation of ethnographic sources. Rather than emphasising the causes of variability in patterns of hunter-gatherer subsistence and settlement systems however, models have been much influenced by the idea of a common structure to all hunting and gathering societies, a structure that can be directly applied to past situations. This supposed structure is often influenced by misplaced analogies used by earlier researchers, or a misreading of ethnographic evidence or interpretations. The ethnographic justification for the supposed common structure of Mesolithic settlement patterns has been a major influence on interpretations of Mesolithic societies.

Price (1973) remarked on the structure of the ethnographically recorded settlement pattern. He noted that in a densely forested environment with widely dispersed resources, the size of the 'subsistence units' was small. Additionally he remarked that people were 'very mobile', never staying in one location for long but 'moving to where resources were available in a seasonal cycle' (Price 1973:

Fishing Large Game Hunting Small Game Hunting Fowlin Table 4.4

Mistassini Cree

Attawapiskat Cree

Round Lake Ojibwa

26 65

39 18

26 53

5

12

16

4

31

5

The idea of a common structure to hunting and gathering societies was much influenced by the concept of the 'foraging adaptation' put forward at the 'Man the Hunter' symposium in Chicago in 1966. The common 'foraging adaptation' included characteristics such as small group size (25-50 individuals), separate male and female foraging patterns (with females providing the bulk of the subsistence

Contributions of groups of resources to diet for Canadian boreal forest groups (Price 1973 after Rogers 1966).

64

ECOLOGICAL AND ETHNOGRAPHIC ANALOGIES

base) and generally low population densities (well below the carrying capacity of the environment). Many of these supposedly common elements were drawn heavily from Lee's research on the !Kung (Lee and deVore 1968; Lee 1979). However it was the concept of a definable and predictable set of behaviours characterising hunter-gatherer populations, rather than the details, which influenced models applied to the Mesolithic. The common model of Mesolithic settlement was in fact much more specific.

MODELS OF INLAND GROUPS

The idea of a definable set of hunter-gatherer behaviour was first taken up by Clark (1972). The most important common element of hunter-gatherer societies for Clark was the predictable seasonal round. This seasonal round in recorded hunter-gatherer groups would be geared towards exploiting resources where and when they were available and thus would be in theory predictable from basic environmental contrasts. In particular Clark was concerned with a possible simplification of this pattern into two main seasons, winter and summer, and two main phases, aggregation and dispersal. On the basis of red deer ecology, Clark (1972), as noted previously, suggested that Mesolithic population would aggregate in the lowlands in winter and disperse to hunt red deer in the uplands in summer (tied to red deer concentrations). Archaeological evidence, particularly the idea that Star Carr was a winter base camp, also further appeared to support the idea of a complementary contrast between upland hunting camps (supposedly occupied in summer) and lowland base camps (supposedly occupied in winter). Although there have been changes in ideas about subsistence and the precise nature of settlement patterns, these contrasting elements continue to structure our understanding of the Mesolithic (figure 4.7), at least partly because, despite changes in specific interpretations, other authors have appeared to find similar evidence for the same basic two-fold contrasts elsewhere.

SUMMER

◄ DISPERSAL ~

Figure 4.7

WINTER

►

--

AGGREGATION

The accepted structure to Mesolithic settlement.

Perhaps the clearest support for the upland hunting site / lowland base camp contrast came from Mellars' (1976) analysis of the functional components of upland and lowland assemblages. Mellars' interpretations were also based on a context of ethnographic and environmental evidence which was used to set up a model of expectations of Mesolithic settlement systems. Like Clark (1972) Mellars suggested that populations would aggregate or concentrate where resources were concentrated, drawing on ethnographic records of seasonal rounds, after Birdsell (1968). Similarly, Mellars suggested that populations would be expected to aggregate in winter in the lowlands where herd animals aggregated. Here long term base camps would also provide protection from predators and allow groups to share food. Again like Clark, Mellars also suggested that the uplands would be occupied by small dispersed hunting groups. Mellars (1976) however addressed the archaeological evidence for these activities by exploring the patterning of site size and assemblage diversity (of the retouched tool component) between upland and lowland sites in Mesolithic Britain. Having identified a

65

POSSIBLE ACTIVITIES

------t1t,11o1:t11=

UPLAND

LOWLAND

THE SPATIAL LOCATION OF SITES

t

t

Two 'ends' of a seasonal system, with few intermediate sites

Figure 4.8

'small' upland sites

'large' lowland sites

locations of sites fitting with 'hunting site' interpretation

few mid-elevation 'neither hunting nor base camp'sites

Archaeological evidence for Mesolithic settlement patterns.

prominent division between large lowland sites either with 'balanced' assemblages (type B sites) or assemblages dominated by scrapers (type C sites), and small upland sites dominated by microliths (typically seen as hunting implements) (type A assemblages), Mellars interpreted the former as base camps and the latter as hunting camps.

subsistence, noted in chapter two, has been important. For another, both the distributions of Mesolithic sites in clear upland and lowland settings (discussed in chapter two) and the work of other authors on assemblage components also highlighted functional contrasts between upland and lowland sites. Jacobi (1978: 320), for example notes the lack of burins on upland sites in the North York Moors compared to many burins found at lowland Star Carr, seeing the latter as clearly a 'domestic' site and the former as 'hunting sites'. Simmons (1979: 112-113) even commented that considering the retouched tool component on sites:

The idea of a functional contrast between upland and lowland sites has remained an important element in interpretations of Mesolithic sites since Mellars' article. For one thing, the concept of a large game hunting phase to 66

Jan

Feb

Mar

Apr

May

Jun

JJI

AuQ

Se::i

Oct

'where only microliths are found, most workers have assumed that they are examining a 'hunting camp' ... by contrast, where the microlith: scraper ratio is more nearly equal then a longer period of settlement with other economic and purely social activities is postulated'.

Although, since Clark's model, the seasonality of 'base camps' has taken on a broader meaning, with some lowland sites being seen as 'summer base camps' - the lowland 'bases' for upland hunting parties - potential summer base camps are still seen as a less permanent occupation than winter base camps (Jacobi 1978; Myers 1986; Simmons 1996). In fact, though interpretations of the location of specific base camps and hunting camps vary, in all interpretations of Mesolithic settlement - from Myers (1986; 1989) to Smith (1992), Spratt (1993) and Simmons (1996) the idea of upland hunting and lowland base camps has remained a strong structural principle.

Fur

Tmiwd srnlpi

······· .............. CBi1tilJ····........ , ....... ,....... .

Seals

.:-'.':'""""''"·"'"

Whfllos

Tomccici

Seals

i

Several authors have attempted to identify the specific ends of a seasonal round - the location of winter base camps and summer sites on the basis of archaeological evidence. Jacobi (1976; 1978) and Myers (1986) thus used raw material sources and common assemblage characteristics to suggest that the wintering camps for Early Mesolithic hunting groups in the Pennines would have been on the Lincolnshire Wolds (discussed in chapter two). Myers (1986) does however alter the model somewhat by suggesting that the exploitation of upland game would have occurred in autumn, prior to a winter time of scarcity. Both authors suggest however that, since raw materials are dominantly derived from local sources in the Late Mesolithic, the settlement pattern would have been more localised at this time (with territory sizes reducing as population densities in contrast increased). The apparent archaeological evidence for lowland base camps and upland hunting camps, and furthermore for distinct 'ends' of a seasonal settlement system have provided substantial support for the two season model of settlement behaviour. This evidence, derived from raw material sources (in the Early Mesolithic) and contrasts in upland and lowlands site sizes and assemblages, is summarised in figure 4.8 (and described in detail in chapter two).

Hucho

Figure 4.9

Resource availability for maritime hunter-gatherers (Rowley-Conwy 1986).

mid 1980s onwards, coastal hunter-gatherers began to take on a new importance, beginning to be seen as a 'special case' requiring a specific model of resource exploitation and settlement. COASTAL COMPLEXITY MODELS

A 'new model' of hunter-gatherer societies emerged in the mid 1980s in contrast to 'the foraging adaptation' (Whitelaw 1990). This new model was inspired by ethnographic records of 'complex' coastal hunter-gatherers, particularly those of the north-west coast of North America, who, living in large permanent or semi-permanent groups failed to comply with the idea of 'simple' hunter-gatherers. The wealth of yearround coastal resources which these societies exploit has been a major focus of attention (Price and Brown 1985; Keeley 1988; Rowley-Conwy 1986, figure 4.9). Marine resources in particular are supposedly the key to the emergence of social complexity in coastal zones. Perlmann (1980), for example, proposes that they hold a unique capacity to support dense populations without agriculture.

Some authors have even applied the model where archaeological evidence is more ambiguous, with apparent support from ethnographic sources. Simmons (1975; 1979; Simmons et al. 1981; 1993), suggests that for the North York Moors, winter base camps would have been on the coast, (with summer hunting in the uplands) - a pattern also maintained in his later (1996) model. Simmons (1979) however also suggests spring and autumn occupation of sites near salmon runs (not included in the later model), an addition partly derived from ethnographic analogies. He comments that:

Very little is known about now-submerged coastal sites or the use of coastal resources in England, especially in contrast to those in Scandinavia (which were uplifted in the early Holocene). The only surviving information in England comes from some Late Mesolithic coastal sites in the extreme west and a few sites in the north-east from the same period. Far from all coastal locations with abundant resources worldwide support 'complex' societies and the exact relationship between resources and complexity is still a major issue. Even if the potential for year-round resources at coastal sites existed (which was not necessarily the case),

'An annual round involving summer hunting on the upland, winter strand-looping, and passing through the intermediate sites twice ... has parallels among groups of recent and near recentfoodcollectors' (Simmons 1975: 9).

The structure of inland settlement at least, thus appears simple and clear-cut. But what about the coast? From the 67

there is remarkably little evidence for either sedentism or complexity in these locations.

LIMITATIONS TO CURRENT ETHNOGRAPHIC MODELS OF SUBSISTENCE AND

In the wake of ideas about coastal complex societies, some authors have nonetheless suggested the presence of sedentary communities, however none of the arguments are convincing. In southern England, Palmer (1980: 439) proposed year-round occupation of Culver Well on the Isle of Portland on the basis of abundant marine resources, however the only support for this notion was that a large shell midden at this site overlying limestone slabs and a cooking pit apparently give the 'appearance of stability' (Palmer 1980: 439). Jacobi (1987: 165) even proposed that across the south-west Peninsula, sedentary communities would have existed where oysters and seals may have filled the winter 'resource gap' (though such sites remain undiscovered). Likewise, though Bonsall (1981: 466) demonstrated the potential for sedentary communities in north-west England, constructing a year-round seasonal resource use schedule for the Eskmeals area of Cumbria, (figure 4.5), there is no clear evidence for sedentary or complex societies at this site. Even where Late Mesolithic coastal sites do exist in some numbers, such as on the western coasts and islands of Scotland, their interpretation is remarkably difficult. Though Ohanian shell middens exist at coastal locations and appear to be distinct from inland microlith-dominated assemblages (Woodman 1989), few would suggest that there is any evidence for sedentism or complexity at these sites. Coastal resources are clearly an important component of any model, but any evidence put forward for a model of coastal complexity remains unconvincing.

SETTLEMENT The application of general models of hunter-gatherer behaviour to societies in Mesolithic Britain has largely taken the form of a 'general model' of settlement (illustrated in figure 4.7). This model, in which hunter-gatherers aggregate in the lowlands at base camps in winter and disperse to upland hunting sites in summer has its origins in Clark's settlement model. A second 'model', that of sedentary complex hunter-gatherers at coastal locations (where marine and terrestrial resources contribute to year-round resource availability) has been less influential, largely contributing rather more to the vague idea that coastal resources were potentially important. Both models appear to have a firm basis in ethnographically documented societies. However in reality this is far from the case. Apart from the simplified use of ethnographic evidence, there are other fundamental problems with models of hunter-gatherer behaviour. These have largely arisen from misunderstandings of ethnographic sources or misplaced analogies, perpetuated by apparent archaeological support. The models used also place constraints on archaeological interpretations by portraying a very static model of huntergatherer settlement, which fails to take into account either short or long-term changes. In fact, particularly for interpretations of inland sites, supposedly ethnographically documented models could be argued to have done more to cloud the issue of Mesolithic settlement than to reveal it. Limitations to the 'inland' model, and to the 'coastal complexity' model, are considered in tum, followed by a discussion of some of the fundamental problems which unite models derived from ethnographic sources.

Seasonal Aggregation and Dispersal The idea of seasonal aggregation and dispersal patterns of hunter-gatherers has clearly had a major influence on models of Mesolithic settlement with Clark (1972) and later Mellars (1976) proposing that groups would have over-wintered at large lowland base camps. Of course, Clark and Mellars based their concept of aggregation on the idea that red deer would concentrate in the lowlands in winter and provide a vital resource. It now seems more likely that red deer lived in relatively small herds (Rowley-Conwy and Legge 1988; 1989) and moreover contributed only a part of Mesolithic subsistence resources (as discussed in chapter three). The idea of long-term occupation of winter base camps has perpetuated however, partly because of ethnographic accounts of long-term sites (such as those occupied by the boreal hunters which Price (1973) and Jochim (1976) considered in detail), and possibly also because if our own concepts of being less mobile in harsh winter weather. The idea of aggregation and long term occupation at winter base camps in the Mesolithic is problematic. As noted in chapter two, the apparent evidence for two 'ends' of a seasonal settlement system in northern England was most probably a 'false pattern' created by a series of biases

68

ECOLOGICAL AND ETHNOGRAPHIC ANALOGIES

affecting the recovery of sites. In any case, the seasonal availability of resources (discussed in chapter three) and of recorded hunter-gatherer exploitation is much more complex than any two-seasonal model. For another, in the highly seasonal environment of temperate Europe the winter is a period of scarcity (Rowley-Conwy and Zvelebil 1989) with fewer resources for long term occupation available than at any other time. Longer term occupation of winter camps is unlikely to have been possible without using stored food, in fact it is only the availability of stored foods that allows boreal groups (such as the Cree, Tanner 1979) to spend the winter in long-term camps. Storage is certainly a possibility for inland Mesolithic groups (Rowley-Conwy and Zvelebil 1989) but it is rarely considered, and has certainly not been a component of the two season model. Even if long-term occupation of winter camps was made possible through the use of stored foods however long term occupation and aggregation are separate issues. Ethnographically documented hunter-gatherer groups aggregate (to maintain wider contacts than the normal coresident group) only at times and places where natural resources are particularly abundant - perhaps the salmon runs mentioned in Simmons (1979) model - and even then rarely for long periods. The boreal hunter-gatherers studied by Price (1973) occupied separate long-term winter sites and short-term aggregation sites in spring when resources were plentiful. The potential distinction at 'large' archaeological sites (interpreted as 'base camps') between long-term occupation and occupation by a larger group is one that is rarely highlighted, although it is clearly very significant. Any 'ideal' model of Mesolithic settlement ought, if an 'ideal' model is even an appropriate tool to use, to incorporate distinctions between aggregation sites and long-term occupation sites as well as a more realistic seasonal separation than that simply between summer and winter occupation. More appropriate means of classifying archaeological sites to relate to ethnographically recorded activities than by the traditional 'base camp' / 'hunting camp' divide would also be vital.

'Base Camps' and 'Hunting Camps' The base camps/hunting camps distinction in fact bears little relationship to ethnographically documented settlement patterns. As well as longer-term occupation sites and aggregation sites, ethnographically documented huntergatherers use different seasonal and task specific sites as well as sites occupied by different members of a co-resident group (which may include all female as well as all-male overnight camps) (Whitelaw 1990). Ethnographic studies (such as those cited by Price (1973) or those of Binford (1978)) emphasise a diversity of site types - for example, large group aggregation sites, short and long term residential camps, specialist exploitation camps for specific resources (such as salmon) as well as hunting 'blinds', short term hunting camps, kill sites and butchery sites. It has been a misreading of ethnographic interpretations, a reliance on interpretations of hunting of large game animals as the subsistence staple (discussed in chapter three) as well supposed evidence for two distinct types of sites in the archaeological record (discussed in chapter two) which has perpetuated the base camp/ hunting camp divide. The nearest to an ethnographic basis for the suggestion of only two types of sites is Binford's (1980) discussion of 'forager' and 'collector' settlement systems. Binford essentially described 'foragers' as mapping onto resources, with residential moves linked to where and when resources were available, whilst 'collectors' would minimise groups movements through planning ahead and making use of storage facilities. He suggested that 'foragers' would leave fewer distinct site types than 'collectors' (that is mainly base camps, extraction camps and aggregation sites). Binford clearly did summarise one type of settlement pattern into three (although not two) types of sites. However, Binford also clearly envisaged the two types of foraging and mobility strategies as a continuum rather than as two distinct patterns that would characterise all hunter-gatherers, past and present and also his 'extraction sites' cover many different activities, far more than any concept of a 'hunting camp'. 'Forager' and 'collector' models were meant to be used as a means of understanding variability in recorded settlement patterns, rather than as a 'blanket model' for past settlement. Whilst ethnographically documented evidence does not support the idea of two site types, aside from the distributions of sites (addressed in chapter two) the archaeological evidence for differences in assemblage characteristics in contrast certainly appears to be suggestive. The idea of lowland winter base camps and upland summer hunting camps is actually supported by several factors - both differences in assemblage constituents (the microlith : scraper ratio) and diversity between the two zones, and also by the relative size of sites (Mellars 1976). All these distinctions are nevertheless problematic. The most obvious limitation is that any distinction between only two tool types (microliths and scrapers) will tend to oversimplify assemblages into two types regardless of other variations. As well as this, changes in the use of these tools through time is another potential problem. Myers (1987) notes that microliths appear to have been used somewhat differently from the Early to the Late Mesolithic (there are more microliths in each haft in the latter period and thus a higher proportion of microliths expected to be discarded and

dominated by microliths, since use wear evidence dominantly represents the use of plant sources, activities at Thatcham may have concentrated on the exploitation of vegetable resources. Simmons (1996) suggests an alternative use of upland sites in the Late Mesolithic by groups clearing and managing upland woodland rather than explicitly hunting, although how these can be differentiated archaeologically is not clear. Further careful excavations and analyses may provide more answers. Detailed excavation of a series of Late Mesolithic sites in the Central Pennines (Spikins 1994; 1995b; 1996a) has revealed that although assemblages were dominated by microliths, a variety of different activities seemed to have taken place; five clearly defined hearths appeared to have been constructed very differently and apparently served different functions. Nonetheless, although new evidence and interpretations are starting to challenge the traditional interpretations, the concept of two basic site types has proved 'hard to shake' influencing both ideas of change through time and limiting explorations of the differences between long-term occupation and repeated use.

preserved in the archaeological record). The use of scrapers also appears to change through time. Though frequent in Early Mesolithic assemblages, scrapers are rare on any recorded Late Mesolithic sites (at March Hill cores have frequently been used as scrapers, and it is not unreasonable to suggest that in more general terms Late Mesolithic cores may also have partly taken over the functions of earlier scrapers). The relative percentages of microliths and scrapers are clearly a poor index of site function given that the use of both (and their relative contributions to assemblages) changes through the period. As well as assemblage composition, the contrast in site size between the uplands and the lowlands is also open to debate. It was demonstrated in chapter two that the nature of upland environments and excavations probably limits the recorded size of upland sites, plus site size is expected to relate to both group size and frequency of reoccupation, which may not be related factors. The base camp / hunting camp distinction is certainly not clearly supported by the archaeological evidence since all the key factors, from assemblage diversity and composition to site sizes, are problematic.

Unlike the 'inland model', the lack of archaeological evidence for coastal sites has made the coastal complexity model somewhat insecure from the outset.

The effects of these biases in 'eroding' our apparently clear record of seasonal base camps and hunting camps is shown in figure 4.10. More than being just an oversimplification of the evidence, the use of contrasting ratios of these two tool types may be hiding differences in settlement structure through time and obscuring much diversity within upland assemblages. Myers (1987) demonstrated that Mesolithic assemblage types were actually divided into more complex categories than the basic groups defined by Mellars (1976), with a series of different assemblage types crossing upland-lowland boundaries. Moreover, Finlayson and Edwards (1997) note that in Scotland, microliths are dominant in all Late Mesolithic 'narrow blade' assemblages regardless of their location (and thus all sites are, strictly speaking, 'hunting camps'). This anomaly is likely to be a function of the rise of microliths (and drop in scrapers) from the Early to the Late Mesolithic, making microliths much more likely to be dominant on Late Mesolithic sites. The lack of any sites which could be interpreted as Late Mesolithic 'base camps' in northern England (although several potential such Early Mesolithic sites exist) may also be explained by the later dominance of microliths amongst retouched tools, with almost all Late Mesolithic sites in this region effectively already having been classified as 'hunting sites' for some time. Aside from changes in tooluse, differences in assemblage diversity between uplands and lowlands can also be affected by sample size, with more apparently 'diverse' assemblages a natural consequence of a larger number of artefacts. Though a rigid division into 'base camps' and 'hunting camps' is not supported by archaeological evidence, or substantiated by ethnographic research it is only recently, particularly as other functions for microliths have been determined (Woodman 1985b; Finlayson 1990a; 1990b; Mithen et al. 1992; Finlayson et al. 1996), that other types of site have been suggested. Healy et al. (1992: 58) for example, suggested that although the assemblage at Thatcham is 70

UPLAND

LOWLAND

THE SPATIAL LOCATION OF SITES

Artefacts recovered most easily where there is upland peat erosion or

Preferential artefact visibility

lowland chalk soils

Low artefact visibility

Artefacts only rarely exposed by erosion or ploughing

Very limited artefact visibility

Very few artefacts recovered on mid elevation pastoral farms or lowland . a IIuv,um

'

(/:,

.//.t

'//'

\,:,

.-:/

,-

.

//,

•;•,-·;,:/-//·,,_· ·'

'/.,

//,

>

,·

/.-~··,; ';.,} 'jW''

·/

/.

./,'I//,

,,.

·/

///:/,1';.,

/ ,,.

/,

•·-'

/

·//,'.

·,: 'c·

/////

/,'

,:,.,.,

,·

·;.

•/,·

//,·

////

/,:::