The Pennsylvania State University

The Graduate School

College of the Liberal Arts

HOUSEHOLD STRUCTURE, DYNAMICS, AND ECONOMICS IN A PREINDUSTRIAL

FARMING POPULATION: THE NORTHERN ORKNEY ISLANDS,

SCOTLAND, 1851-1901

A Dissertation in

Anthropology and Demography

by

Julia Anne Jennings

2010 Julia Anne Jennings

Submitted in Partial Fulfillment of the Requirements for the Degree of

Doctor of Philosophy

December 2010

The dissertation of Julia A. Jennings was reviewed and approved* by the following:

James W. Wood Professor of Anthropology and Demography Dissertation Advisor Chair of Committee

Patricia L. Johnson Associate Professor of Anthropology, Demography, and Women‘s Studies

Stephen A. Matthews Associate Professor of Sociology, Anthropology, Demography, and Geography

Kenneth G. Hirth Professor of Anthropology

Duane F. Alwin McCourtney Professor of Sociology and Demography

Debashis Ghosh Associate Professor of Statistics

Nina G. Jablonski Professor of Anthropology Head of the Department of Anthropology

*Signatures are on file in the Graduate School

iii ABSTRACT

This dissertation examines the demography, economy, and structure of agricultural households in the Northern Orkney Islands, Scotland, in the last half of the nineteenth century

(1851-1901). The household is used as the unit of social, demographic, and economic analysis throughout this thesis. Household structure in Orkney is described using both case studies and aggregate data. From this information, the compound household, which would go unnoticed if only documentary evidence were used, is defined, described, and integrated into existing systems of household classification. Patterns of household types found in Orkney are compared to published accounts of historical populations throughout Europe. Household dynamics are explored using event-history models of the transition between household types. Finally, the household as an economic unit is examined in a study of the hiring of life-cycle servants, both generally and with respect to the sex of the servant.

Multiple sources of information, about both the physical farmsteads and the people who occupied them, show that households in Northern Orkney are more structurally and genealogically complex than has been previously reported for Northwest Europe in general.

These findings could not have been made without the use of data sources (e.g. archaeological remains) that supplement documentary evidence such as census listings. It may be the case that the kinds of data commonly used by historical demographers have actually underestimated the degree of household complexity in the preindustrial past, as household complexity may actually be ―hidden‖ in these records. Event-history models of transitions between household types demonstrate that compound and extended households (characterized by the presence of multiple dwellings and non-nuclear kin, respectively) have rather different dynamic tendencies. Finally, analyses of the presence of life-cycle servants show that households are adaptive economic units

iv that can respond to labor shortages (both general and sex-specific) through the hiring of life-cycle servants.

v TABLE OF CONTENTS

LIST OF FIGURES ...... vii

LIST OF TABLES ...... xi

ACKNOWLEDGEMENTS ...... xiv

Chapter 1 Introduction ...... 1

The Household as a Unit of Analysis ...... 2 Definition of Households ...... 4 Data and Setting ...... 8 The Economy of Northern Orkney, 1851-1901: A Period of Change ...... 12 Outline ...... 14 Tables and Figures ...... 16 References ...... 19

Chapter 2 Households in Northern Orkney, 1851-1901 ...... 23

Introduction ...... 23 Background ...... 24 Household Histories using Multiple Data Sources: Case Studies ...... 31 Households in Orkney: A Descriptive Overview ...... 43 Discussion ...... 48 Figures and Tables ...... 52 References ...... 107

Chapter 3 Dissolution and Formation of Extended-Family and Compound Households in Northern Orkney, Scotland, 1851-1901 ...... 111

Introduction ...... 111 Background ...... 112 Data and Methods ...... 125 Results ...... 128 Discussion and Future Directions ...... 132 Tables and Figures ...... 134 References ...... 143

Chapter 4 Household Predictors of the Presence of Life-cycle Servants in Northern Orkney, Scotland, 1851-1901 ...... 149

Introduction ...... 149 Background ...... 152 Data, Methods, and Results ...... 169 Household-level Predictors of the Number of Life-cycle Servants: Methods ...... 171 Household-level Predictors of the Number of Life-cycle Servants: Results ...... 172

vi Household-level Predictors of the Number of Life-cycle Servants by Sex: Methods ...... 173 Household-level Predictors of the Number of Life-cycle Servants by Sex: Results ...... 175 A Comparison of the Source and Sink Households of Life-cycle Servants ...... 176 Discussion ...... 178 Tables and Figures ...... 183 References ...... 196

Chapter 5 Discussion and Future Directions ...... 201

Discussion ...... 202 Future Directions ...... 206 References ...... 209

Appendix A Valuation and Acres of Arable and Pasture ...... 211

Appendix B Measuring Consumer-to-Producer Ratios: Do Weighting Systems Matter? ...... 218

Appendix C A Contemporary Description of Servants and Hiring Fairs ...... 237

Appendix D An Example Life History of a Farm Servant ...... 239

References ...... 240

vii

LIST OF FIGURES

Figure 1-1: Satellite image showing the location of the Orkney Islands...... 16

Figure 1-2: Satellite image showing the location of the six study islands...... 17

Figure 1-3: Histogram of reported ages, 1901 Census returns...... 18

Figure 2-1: Genealogical relationships among the inhabitants of South Hammer...... 54

Figure 2-2: Historical and modern maps of South Hammer with building types labeled...... 55

Figure 2-3: Modern photograph of the South Hammer croft complex ...... 56

Figure 2-4: Genealogical relationships among the inhabitants of Sangar...... 58

Figure 2-5: Historical and modern maps of Sangar with building types labeled...... 59

Figure 2-6: Modern photograph of the Sangar croft complex ...... 60

Figure 2-7: Genealogical relationships among the inhabitants of Quoynabreckan...... 62

Figure 2-8: Historical and modern maps of Quoynabreckan with building types labeled...... 63

Figure 2-9: Modern photograph of the Quoynabreckan croft complex ...... 64

Figure 2-10: Genealogical relationships among the inhabitants of Ha‘Ouse...... 66

Figure 2-11: Historical and modern maps of Ha‘Ouse with building types labeled...... 67

Figure 2-12: Modern photograph of the Ha‘Ouse croft complex ...... 68

Figure 2-13: Genealogical relationships among the inhabitants of Stanegeo...... 70

Figure 2-14: Historical and modern maps of Stanegeo with building types labeled...... 71

Figure 2-15: Modern photograph of the Stanegeo croft complex ...... 72

Figure 2-16: Genealogical relationships among the inhabitants of Nether Brough...... 75

Figure 2-17: Historical and modern maps of Nether Brough with building types labeled...... 76

Figure 2-18: Modern photograph of the Nether Brough croft complex, southern block ...... 77

Figure 2-19: Modern photograph of the Nether Brough croft complex, northern block ...... 77

viii

Figure 2-20: Genealogical relationships among the inhabitants of Moa...... 80

Figure 2-21: Historical and modern maps of Moa...... 81

Figure 2-22: Modern photograph of Moa ...... 82

Figure 2-23: Percentage of sample by age and relationship to the household head, males 1851 ...... 84

Figure 2-24: Percentage of sample by age and relationship to the household head, females 1851...... 84

Figure 2-25: Percentage of sample by age and relationship to the household head, males 1861...... 85

Figure 2-26: Percentage of sample by age and relationship to the household head, females 1861...... 85

Figure 2-27: Percentage of sample by age and relationship to the household head, males 1871...... 86

Figure 2-28: Percentage of sample by age and relationship to the household head, females 1871...... 86

Figure 2-29: Percentage of sample by age and relationship to the household head, males 1881...... 87

Figure 2-30: Percentage of sample by age and relationship to the household head, females 1881...... 87

Figure 2-31: Percentage of sample by age and relationship to the household head, males 1891 ...... 88

Figure 2-32: Percentage of sample by age and relationship to the household head, females 1891 ...... 88

Figure 2-33: Percentage of sample by age and relationship to the household head, males 1901 ...... 89

Figure 2-34: Percentage of sample by age and relationship to the household head, females 1901...... 89

Figure 2-35: Headship rates by age and sex of head and census year...... 90

Figure 2-36: Percentage of sample living in simple households by age and census year ...... 93

Figure 2-37: Percentage of sample living in extended households by age and census year ... 94

ix Figure 2-38: Percentage of population living in compound households by age and census year ...... 95

Figure 2-39: Histogram of valuation, 1881 ...... 97

Figure 3-1: Genealogical relationships of the inhabitants of South Hammer, 1901...... 135

Figure 3-2: Plan of the surviving structures at South Hammer, 2003...... 136

Figure 3-3: OSGB map of South Hammer, showing farm and dwelling structures, 1901 ...... 137

Figure 3-4: Hazard of household dissolution by census year...... 138

Figure 3-5: Hazard of household formation by census year...... 138

Figure 4-1: Chayanov‘s model of household cycles in C/P ratio...... 183

Figure 4-2: Histograms of the age distribution of male servants by census year...... 186

Figure 4-3: Histograms of the age distribution of female servants by census year ...... 186

Figure 4-4: Histograms of the age distribution of the sample population by census year...... 187

Figure 4-5: Histogram of the C/P ratio of the source household when the future servant was last observed...... 193

Figure 4-6: Histogram of the C/P ratio of the sink household before the servant arrives...... 193

Figure 4-7: Histogram of the C/P ratio of the source household in the census interval after servant leaves...... 194

Figure 4-8: Histogram of the C/P ratio of the sink household with the servant observed in it ...... 194

Figure 4-9: Distribution of the difference in C/P ratio between source and sink houseohlds ...... 195

Figure 4-10: Distribution of the difference in valuation between source and sink households...... 195

Figure A-1: Actual versus predicted valuation...... 216

Figure B-1: Chayanov‘s model of household cycles in C/P ratio...... 225

Figure B-2: Observational study of weighted contribution to production and consumption by age and sex...... 226

Figure B-3: Physiological work capacity by age and sex...... 227

x Figure B-4: Profiles of male production weights by age ...... 228

Figure B-5: Profiles of female production weights by age ...... 228

Figure B-6: Profiles of male consumption weights by age...... 229

Figure B-7: Profiles of female comsumption weights by age...... 229

Figure B-8: Scatterplots of C/P measurements...... 231

Figure B-9: Coefficients of models of demographic outcomes ...... 232

Figure B-10: Coefficients of models of household-level outcomes...... 233

Figure B-11: Coefficients of models of economic outcomes...... 233

xi

LIST OF TABLES

Table 1-1: Population of the Northern Orkney Islands, 1851-1901...... 18

Table 2-1: Percentage of households by type and census year ...... 52

Table 2-2: The inhabitants of South Hammer by census year...... 52

Table 2-3: The inhabitants of Sangar by census year...... 57

Table 2-4: The inhabitants of Quoynabreckan by census year...... 61

Table 2-5: The inhabitants of Ha‘Ouse by census year...... 65

Table 2-6: The inhabitants of Stanegeo by census year...... 69

Table 2-7: The inhabitants of Nether Brough by census year...... 73

Table 2-8: The inhabitants of Moa by census year...... 78

Table 2-9: Percentage of population by household kin status relative to the head...... 83

Table 2-10: Percentage of households that contain kin ...... 83

Table 2-11: Mean household size by type and census year...... 96

Table 2-12: Mean valuation by household type and census year...... 96

Table 2-13: Median valuation by household type and census year...... 96

Table 2-14: Composition of households in 1851 using Hammel-Laslett system...... 98

Table 2-15: Composition of households in 1861 using Hammel-Laslett system ...... 99

Table 2-16: Composition of households in 1871 using Hammel-Laslett system ...... 100

Table 2-17: Composition of households in 1881 using Hammel-Laslett system...... 101

Table 2-18: Composition of households in 1891 using Hammel-Laslett system...... 102

Table 2-19: Composition of households in 1901 using Hammel-Laslett system...... 103

Table 2-20: Comparative proportions of family types ...... 104

xii Table 2-21: Percentages of extended, multiple and compound households, Orkney and other European populations...... 105

Table 3-1: Number and proportion of household types by census year...... 134

Table 3-2: Inhabitants of South Hammer, 1901...... 134

Table 3-3: Weighting system for consumer-producer ratios ...... 139

Table 3-4: Composition of simple family households by census year ...... 139

Table 3-5: Extended family households by type of extension and census year...... 139

Table 3-6: Composition of compound households by census year...... 139

Table 3-7: Average household size by type and census year...... 140

Table 3-8: Average number of vital events per census interval by household type...... 140

Table 3-9: Mean valuation by household type ...... 140

Table 3-10: Average and standard deviation of consumer-producer ratios by household type ...... 140

Table 3-11: Number of households by ownership status and type...... 141

Table 3-12: Esimates of covariate effect on the dissolution of extended and compound households, discrete time logit hazard models...... 141

Table 3-13:Estimates of covariate effects on the formation of extended and compound households, discrete time logit hazard models...... 142

Table 4-1: Chayanov‘s model of changing household composition ...... 183

Table 4-2: Weighting system used to calculate C/P ratios ...... 184

Table 4-3: Birthplaces of servants, 1851-1901...... 184

Table 4-4: Total sample population by census year...... 184

Table 4-5: Sample population between ages 12 and 30...... 184

Table 4-6: Number and mean age of servants by sex and census year ...... 185

Table 4-7: Distribution of individual age differences between spouses...... 185

Table 4-8: Mean values of the household-level predictor variables for households with no servants...... 187

xiii Table 4-9: Mean values of the household-level predictor variables for households with servants...... 188

Table 4-10: Mean values of head-related predictor variables for households with no servants...... 188

Table 4-11: Mean values of head-related predictor variables for households with servants ... 188

Table 4-12: The number of household in the sample by census year and presence of servants...... 189

Table 4-13: Results of Poisson regression models for the number of servants present in the household...... 189

Table 4-14: Mean values of household-level predictor variables for households with at least one male servant...... 190

Table 4-15: Mean values of household-level predictor variables for households with at least one female servant...... 190

Table 4-16: Mean values of head-related predictor variables for households with at least one male servant ...... 190

Table 4-17: Mean values of head-related predictor variables for households with at least one female servant ...... 190

Table 4-18: Results of Poisson regression models for the number of servants by sex...... 191

Table 4-19: Results of Wald tests of the equality of coefficients across equations...... 192

Table 4-20: Distribution of houseohld C/P ratios in the four census records...... 195

Table A-1: Properties included in the analysis...... 215

Table A-2: Results of the multiple linear regression of valuation...... 216

Table B-1: Chayanov‘s model of changing household composition ...... 225

Table B-2: Hammel‘s modification of Chayanov‘s weights used to calculate C/P ratios ...... 226

Table B-3: Energy requirements by age and sex of rural Gambians...... 227

Table B-4: Correlation coefficients for C/P ratios...... 230

Table B-5: Total absolute difference of C/P measurements from mean coefficients for all simple models...... 233

xiv ACKNOWLEDGEMENTS

In preparing this dissertation, I have been the fortunate recipient of assistance, advice, and support from many sources. I would like to thank my mentor and advisor Dr. James Wood for his support, encouragement, and guidance. He has shown constant faith in my abilities, even when I am plagued with doubt and worry (which probably happens more than it should). When we first met for coffee in 2005, I would have never thought that I would take scholarly interest in topics as diverse as unobserved heterogeneity, farmer‘s autobiographies, and (of course) muck.

Dr. Pat Johnson has provided insightful comments on this thesis as well as on matters personal, professional, and egg-related. I am thankful for her generous hospitality, both in Orkney and in

State College, and very practical advice.

This work been improved by the thoughtful input of my dissertation committee. I owe most of my knowledge of spatial analysis to Dr. Stephen Matthews. Dr. Ken Hirth has offered thought-provoking questions about the household economy. Dr. Duane Alwin has great breadth of intellectual and personal interests, a quality I hope to emulate. Dr. Debashis Ghosh has provided indispensible advice on statistical programming and modeling.

I am in great debt to the other members of the North Orkney Population History Project.

Dr. Timothy Murtha provided archaeological survey data as well as a crash-course in buildings archaeology. Dr. Corey Sparks designed the population database and showed me the ropes. He has always been willing to assist when questions arise.

My career in anthropology to date has been shaped by many mentors, advisors, and colleagues. Drs. George Alter, Myron Gutmann, and Susan Leonard invited me to the

Longitudinal Analysis of Historical Demographic Data course at the University of Michigan.

This thesis employs many of the techniques I learned at that course. They have also enthusiastically supported my work in historical demography and introduced me to the family of

xv scholars in the field. My collaborative work with Dr. J. David Hacker has been a fantastic experience and he has provided consistent encouragement and timely professional advice. Dr.

Gordon DeJong (through the Population Research Institute) has provided generous financial support, which allowed me to strengthen my training and experience in demography. I would also like to thank Drs. John Verano, Trent Holliday, and Susan Sheridan who kindly gave of their time and encouraged me as an undergraduate anthropology major. I also owe special thanks to

Dr. Winifred Creamer, my first mentor in anthropology, who remains an inspiration and example of academic excellence.

Many people assisted my data collection efforts in Orkney and Edinburgh. Lucy Gibbon,

Sarah Grieve, Alison Fraser, and David Mackie graciously and patiently guided me through

Orkney‘s archival resources. After long hours in the archives, Lucy was still willing to meet me for a pint at the Bothy Bar or to watch an episode of Dr. Who. The staff of the General Register

Office for Scotland provided access to vital registers for Northern Orkney. Mr. and Mrs. Delday of Lerona in Kirkwall are fabulous hosts and make the best breakfast in town. The people of

Northern Orkney have been supportive throughout the development and implementation of the

North Orkney Population History Project. They deserve many thanks. I also wish to acknowledge the Westray Heritage Centre and Buildings Preservation Trust, and Billy Brown, the late registrar on Westray, who have supported and assisted many members of the project.

I owe many, many thanks to the wonderful support staff in the department of anthropology. Betty Blair, Kim Miller, Faye Maring, Jasun Lego, Stephanie Rossman, and

Wendy Fultz have offered kind and patient assistance throughout my time at Penn State.

My friends and colleagues at Penn State have been a vital source of support, advice, kindness, and laughter. I am grateful for the friendship of Denise Liberton, Ellen Quillen, and

Nicole Ortmann—you ladies made that first year bearable! Jen Wagner has been a great friend

xvi and fellow strategist during the job search. I have had many interesting talks with Dr. Heath

Anderson (some were even scholarly). Ruscena Wiederholt has been my fellow agent-based model enthusiast. Carolyn Keagle, Jessica Leger, and I have shared some much-needed retail and gastronomic therapy. Drs. Erick Rochette, Craig Goralski, and Dawn Miller have given me great advice on all matters pedagogical. Finally, Paul Ayayee has borne the worst of my worries with grace and humor. I cannot thank you enough.

My family has been a steady source of support. They have always encouraged my academic pursuits and have been very patient and understanding as I complete them. I owe a great debt to Monica, Lamont, and Lauren Jennings and Bob and Joy Hanlon. You‘re the best non co-resident kin group!

This research was supported in part by the National Science Foundation (HSD052739), the Eunice Kennedy Shriver National Institute for Child Health and Development (T-

32HD007514, to the Population Research Institute), The Pennsylvania State University

Department of Anthropology, and the Population Research Institute.

Chapter 1

Introduction

In many societies the household is the one unit through which almost the whole of the economy can be studied as a connected and functioning whole. – Rosemary Firth (1966:1).

I believe that a convincing case can be made out in favour of the household as the fundamental unit in pre-industrial European society for social, economic, even educational and political purposes…. The relationship between parents, children, servants and kin within the English household, and the interplay of its size and structure with economic and demographic development, make up an intricate adaptive mechanism which we are only now beginning to understand. – Peter Laslett (1969: 201, 223).

In the small tribal societies which are the traditional subject of anthropological research, direct exploitation of natural resources is usually carried out for the most part by families, and the economic processes of distribution, use, and consumption of goods occur largely within the family context. Thus in most African societies a family group of some kind produces food by cultivating its own land or herding its own cattle and consumes most of this food at its own homestead. Natural resources are major points of ecological contact between human communities and their physical environment, and in tribal societies the ecological processes of exploitation occur mainly in a family milieu. The labour and skills necessary for exploiting the natural resources are funneled through the family in actual application, while the goods consumed are distributed through family channels. Individuals obtain their vital sustenance largely through the mediation of the family structure. Thus the family is a principal locus of ecological processes. – Robert F. Gray (1964:5).

God save me from what they call households. – Emily Dickinson, letter to Abiah Root, 1850.

This thesis examines the demography, economy, and composition of agricultural households in the Northern Orkney Islands, Scotland, in the last half of the nineteenth century

(1851-1901). This task is taken up in three largely independent research papers. While each of the following chapters presents separate research topics, a unifying thread runs through them, namely the importance of the household as a unit of demographic, social, and economic analysis.

This focus on households allows for the study of processes above the level of the individual, but

2 below the level of the population as a whole. In Orkney in this period, households are primarily engaged in traditional farming. Farming households, perhaps more so than industrial households, manage their resources collectively and share the products of their labor. Thus, the household is an important point of interaction between individuals and the environment, and is the context for economic and demographic decision-making and change. However, the task of defining and describing these fundamental social groups is challenging, and is in fact surprisingly difficult given their importance and ubiquity. In this introduction, households are presented as a unit of analysis, conceptual definitions of households are discussed, and an operational definition of households is proposed and will be employed throughout the thesis. Then, the data used in the study are described. Finally, some background regarding economic changes during the period of interest is provided and the structure of the thesis is outlined.1

The Household as a Unit of Analysis

Gene Hammel wrote that a household is ―the next bigger thing on the social map after an individual‖ (1984:40-41). In many anthropological and archaeological studies, households are considered a fundamental unit of social and economic organization (Netting et al. 1984; Wilk

1991; Goody 1976; Santley and Hirth 1993; Ashmore and Wilk 1988). Households are the context in which demographic and subsistence activity occur. They are concerned primarily with survival and reproduction and they organize household labor and resources to improve their chances of success in that enterprise. To meet their needs, households engage in a set of activities that include a combination of production, distribution (consumption, exchange, pooling, etc.), transmission (transfer of property), biological and social reproduction, and coresidence (Netting

1 The ecological, demographic, and social setting of the Northern Orkney Islands is not discussed in detail here. Interested readers should refer to Sparks (2007), Brennan (1979), Fenton (1997), and Thomson (2008a; 2008b).

3 1993). In this sense, households are the unit in which a variety of essential behaviors overlap and intersect.

It has been argued that households have emergent qualities, or that they are more than a simple collection of individuals (Netting et al. 1984). While households can be considered a

―knot of individual interests‖ (Laslett 1984), household decisions emerge through the collective action of their members and may include bargaining, disagreement, and conflict. Single individuals who act independently seldom make decisions with important demographic ecological consequences. Rather, households are the context for decisions about marriage, fertility, farm management, and the deployment of resources (including labor, goods, money, etc.). Households are subject to external stresses, including poor harvests, unfavorable prices, and sociopolitical hardships such as warfare, but they are also prone to internal stresses that arise from the normal household life cycle. In his work on the peasant economy, Chayanov (1986) showed how changing household composition affects the degree of stress or the level of production that must be met to maintain subsistence as a newly formed household grows and matures. Thus, in a relatively homogenous society, households may fare differently from one another simply because they are at different stages in their life cycle or are composed of different members. Household composition therefore affects both the level of household needs and the ability of the household to meet those needs, which may have important consequences for the success of the domestic group and the well-being of its members.

Household ―complexity‖ or ―simplicity‖ is, at least in part, a function of household composition, or the personnel that make up the domestic group. Traditionally, historical demographers have defined household complexity in terms of the kinds of kin found within the household (Laslett 1977). If households contain kin additional to a married couple and their children, such as cousins, elderly parents, or grandchildren, historical demographers usually consider call them complex. Archaeologists consider household complexity differently, as they

4 must draw inferences from the physical remains left by these groups. Household complexity in this sense may include categories based on the size and number of buildings associated with a homestead, or evidence of the variety of activities the household engaged in, such as economic production (Hirth 2009; Santley and Hirth 1993; Hirth 1993). Contemporary demographers are able to consider a broader range of human activity that may contribute to household complexity in the behavioral and residential sense. These topics include the role of migrant household members who send home remittances, the rise of cohabitation, and the phenomenon of ―living apart together‖ (Sana and Massey 2005; Igzigsohn 1995; Smock 2000; Bumpass and Raley 1995;

Winfield 1985; Levin 2004). Unfortunately, the data available from Orkney in the period of interest (1851-1901) do not allow for the detailed study of such behaviors, as they seldom (if ever) left traces in the historical record. In this thesis, household complexity combines the perspectives of historical demography and archaeology, as the data available allow for the observation of both kin relationships within the household and the physical structures that these individuals occupied and used.

Definition of Households

Despite the importance and ubiquity of households as units of analysis in the social sciences, commonly accepted conceptual definitions of households have proved difficult to formulate. Both within and between societies, households perform a variety of tasks that overlap to different degrees. In addition, household personnel vary between households and changes occur over time within households. Even if scientists could arrive at a single conceptual definition of a household, this definition may vary from the folk categories that individual members of a society use to define their own domestic groups (Hammel 1984).

5 While there are many conceptual definitions of the household, they usually include at least one or two of the following criteria: location, shared activity (especially production and consumption), and kinship (Laslett 1977; Netting et al. 1984). Often, the family is distinguished from the household using these criteria. In general, households are coresident, either under the same roof or within the same compound or cluster of buildings, but not necessarily related, as households may include servants, boarders, and lodgers in addition to related individuals.

Families in contrast, are related, but not necessarily coresident. Attempts at definition of the household vary from those specific to a particular group (or village, or region, etc.) to more generalized conceptions (for a critique of universal definitions, see Yanagisako 1979).

Wilk and Netting (1984) view the problem of the definition of households in two ways.

The first relates to household morphology and structure, which raises the issue of how boundaries should be drawn around household groups and in what way these groups should be compared to each other both within and between societies. The second problem lies in the behavior of households, which involves the identification of the primary group that performs the functions of production, consumption, reproduction, and so on, in a given society. Many questions in the study of households, and the topics raised in this thesis, relate to both the structure of households and their many functions.

Another problem in the definition of households comes to us through historical precedent. In Western societies, households have been defined for the purposes of population enumeration for centuries. Because of this, households, however conceptualized, were defined such that a person may only belong to one household at a time to avoid multiple enumerations.2

The data from this study, as well as any other work using census-like records, will be colored by the definition of households employed by enumerators in the society of interest. The criteria used

2 These historical definitions took many forms. These include ―eating at one table‖ or ―sharing one cooking pot‖. See Wood (n.d.) for additional discussion.

6 to define households for counting purposes may not directly correspond to what contemporary people considered their household (or households if they identified with more than one group) or the fundamental domestic group of interest to social scientists.

Hammel (1984) provides an interesting abstract definition of households. He suggests that households are ―the largest supraindividual (and perhaps named) group with the greatest multifunctional corporacy‖ (1984:41). This definition has meaning for both the structure of households (how to draw circles around a group) and the function of households (the component of shared activity). Further, the suggestion that these groups might be named is interesting the context of Orkney, where houses and farmsteads are named, so that their inhabitants may be thought of a named group, such as South Hammer, or the inhabitants of South Hammer.3

Additional definitional nuances of households arise in the context of traditional farming societies. In these societies, individuals tend to interact with their environment at the level of the household, as households make decisions regarding land, labor, and resource management.

Households manage agricultural activities, provide the labor needed to perform these activities, and subsist primarily from the products of their work on the land.4 Russian agronomist A. V.

Chayanov linked the farming economy directly with the composition of the farming family. He wrote:

The first fundamental characteristic of the farm economy of the peasant is that it is a family economy. Its whole organization is determined by the size and composition of the peasant family and by the coordination of its consumptive demands with the number of its working hands (1931:144).5

If Chayanov‘s assertion is correct, because people are grouped into households when undertaking essential subsistence activities and making decisions with economic and demographic

3 Farm names are even incorporated into people‘s names because of the small number of surnames. Examples include Willy o‘ Goltiquoy (William Seatter) and Geordie o‘ Trenabie (George Rendall). 4 For a detailed discussion of farming households, see Wood (n.d.). 5 Here, ―household‖ might reasonably be substituted in place of ―family.‖

7 consequences, households and how they vary in size, age, sex, and kin composition, and other characteristics are essential to the study of agrarian population dynamics.

Both the general and agrarian-specific concepts of the household indicate some important characteristics of households that can be measured, evaluated, and used in analysis. These include characteristics of the members of the household, such as age and sex, which partially determine household consumption requirements and labor capacity. The size (number of members) of the household is also potentially important, as larger households generally have larger and more diverse labor pools to draw on, but also more mouths to feed. Kin relationships within a household are indicative of diversity in membership beyond the nuclear family and relate to household ―complexity‖ as understood in terms of household personnel. The amount of land to which the household has access may determine the amount of labor that needs to be deployed by the household, the level and intensity of production, and financial burdens such as rents. Other household characteristics, such as the age (and possibly sex) of the head may indicate a household‘s position in its life cycle, as young heads face different challenges, such as dependent children, than older heads, who may have to adapt to labor shortages as children leave home.

Although the concept of the household and the prospect of a universally applicable definition remain uncertain, if even possible, an operational definition of the household for use in this study will clarify how households are identified and measured for the analyses that follow.

While this definition was specifically developed for this study, it may prove useful when making comparisons to other regions.

In reference to the three-part definition of households (shared activity, location, and kinship), in Orkney location and kinship can be determined through census enumerations and vital records. The study of shared activity in the past is more difficult. Analogies of household behavior can be drawn from studies of other traditional agrarian societies, in which the household is also the major farming unit (Netting 1993; Wood n.d.). Members of the household provided

8 labor inputs for agricultural production (including ―domestic‖ work such as food processing, storage, haulage of fuel, cooking, and distribution), and they could expect to be provided with food, shelter, and other necessities from the pooled output of the farming unit. Thus, in most preindustrial farming systems, the household and farming unit are more or less synonymous.

For the purposes of this study, households are operationally defined as all the inhabitants of a single named farmstead as listed in decennial census enumerations. Evidence from the physical structures present at the farmsteads (discussed in detail in chapter 2), suggest that this grouping meets Hammel‘s (1984) criterion of the largest social group with the greatest corporate behavior, as these farmsteads have a single set of farming structures with associated dwellings and plots of land. When a household shares a single set of essential farming buildings, it is presumed that farming was conducted cooperatively by the group. While this definition is specific to this study and may not be readily generalizable beyond it, the specification of how households are identified for observation and measurement ensures that future work in this region can examine comparable groups of individuals and that varying definitions of households can be considered when making cross-cultural comparisons.

Data and Setting

The Northern Orkney Islands are found to the north of mainland Scotland, where the

North Sea and Atlantic Ocean meet (Figures 1-1 and 1-2). The islands that comprise the study area are Westray, Papa Westray, Sanday, Eday, North Ronaldsay, and Faray. This thesis uses a variety of data sources, some of which are commonly used in historical demographic research, such as census returns and vital registers, while other sources, such as archaeological surveys of abandoned crofts, distinguish this work from traditional approaches to the study of population history.

9 These data were collected as part of an ongoing research effort known as the North

Orkney Population History Project (Murtha et al. 2008). The goals of this project are to investigate population dynamics from the nineteenth century until the present day. Topics of particular interest to the research group include demographic, economic, social, and land use change. This thesis focuses on documentary evidence (census returns, vital registers, valuation rolls, and historical maps) and archaeological surveys, but other information, including ethnographic interviews, oral histories, land quality maps, and staple grain prices, are part of this rich dataset.

Individual-level census returns provide information about individuals, such as age, marital status, and occupation, and households, as these listings indicate how the inhabitants of the islands are grouped into households, and how individuals within households are related to each other. The decennial census returns used in this study come from the years 1851, 1861,

1871, 1881, 1891, and 1901, with the last year determined by United Kingdom regulations that restrict the release of individual-level census data for 100 years. In Orkney, house names are listed in census returns, which allow researchers to identify the location of the house or farmstead occupied by the listed inhabitants.

Vital registration of births, deaths, and marriages, are a second source of demographic information about the Orkney Islands. These records of vital events include information about the individuals involved, such as age, sex, and marital status, and sometimes their immediate relatives (spouse, children, or parents) as well. This information includes the date of the vital event and other relevant details, such as occupation of and place of residence.

Valuation rolls are a source of data about the economic standing of households. These records were collected for taxation purposes, and they provide information about the taxable value of properties, the owner of the property, and their tenants. The valuation of properties is closely associated with both the amount of land (in acres of arable and pasture) to which the

10 occupants have access, as well as the rental value (See Appendix A for details). Thus, valuations can be considered proxy measures of landholding size and quality. These records, which also list farm names, can be linked to their corresponding inhabitants in the census returns, as farm

(house) names were stable over time. Maps predating the study period (pre-1851) show the names and locations of farms that are also present on maps from the study period, such as the

1882 Ordinance Survey, and on modern maps.

Additional information about households comes from historical maps, which detail the footprints of buildings present at farmsteads, and are often annotated by landlords to indicate field boundaries. These maps help in the reconstruction of the physical structures present at a particular farmstead and how they change over time. Archaeological surveys of abandoned crofts, directed by Dr. Timothy Murtha, provide insight into the function of farm buildings

(whether they served as dwellings, barns, byres, and so on), the sequence of construction, and the repurposing of buildings. Traditional and aerial photography of farmsteads supplement archaeological surveys and historical maps of the physical structures in which household members lived and worked.

The North Orkney database includes multiple sources of information about life in the

Northern Isles and provides important historical, social, and environmental contexts that are often unavailable to students of population history. However, the data do have certain weaknesses and inconsistencies. Some of these problems are familiar to historical demographers. These include the bias that is introduced to the dataset through the process of record linkage or family reconstitution. For example, it is highly likely that individuals who can be traced from birth, to marriage, and then to death, are different from individuals for whom complete sets of records cannot be found (Bonneuil 1997; Jonker and van der Vaart 2007; Kasakoff and Adams 1995; Lee

1993; Oeppen 1993; Ruggles 1992; Sieder and Mitterauer 1983; Wrigley 1994; Wrigley et al.

1997). Migration is generally thought to be an important cause of incomplete linkage. Out-

11 migration from Orkney was common in the nineteenth century, but our records do not indicate the identity of migrants or the timing of their moves (Thomson 2008a, 2008b).

In addition, observation of the composition of households, which is essential to the analyses reported in this thesis, is only possible using census records, which are subject to 10- year interval censoring. This censoring makes measurements of household change less exact, as the cause and timing of the changes are unknown. Vital registration can fill in some of the gaps in the census records, as the registers give the exact dates of births, deaths, and marriages.

However, it is probable that vital events, particularly deaths, are under-reported (Jennings and

Wood 2008). Age heaping is another source of inaccuracy and inconsistency in the historical records, particularly the census returns. Histograms of the age of individuals taken from the census records show noticeable age heaping, as either census enumerators or the individuals themselves reported ―rounded‖ ages (Figure 1-3). Other inconsistencies in the data, such as multiple spellings of proper names and errors in the reporting or transcription of dates, are known to exist.

The samples drawn from this database will vary from chapter to chapter. This variation reflects the current state of record linkage and the information needed to address the research question at hand. In all chapters, households are linked using their farm or house names. This allows households to be tracked from census year to census year and the vital events that take place in a particular household to be identified. However, not all households can be linked in this way, which affects both sample size and composition, and may be a source of sampling bias. In particular, the residents of villages, such as Pierowall, Westray, cannot be linked using house names, as households within the village are difficult to distinguish from each other and track over time with the current state of record linkage. The details of the sample selection criteria will be outlined in each chapter.

12 The Economy of Northern Orkney, 1851-1901: A Period of Change

The period of study, 1851-1901, covers a time of economic and social change to which households needed to adapt. Traditionally, the Northern Orkney Islands relied on smallholder agriculture, with emphasis on grain crops, particularly oats and bere (a landrace of barley).

Subsistence agriculture was supplemented by animal husbandry, fishing, and craft production.

This traditional system changed during the ―kelp boom‖, which began in 1780 and lasted for approximately 50 years (Thomson 1983). Kelp is a product made from burned seaweed (not the seaweed itself) used to produce alkali for glass, soap, and dye manufacturing. The population of the Northern Islands, particularly Westray, Papa Westray, Sanday, and North Ronaldsay, which were the parts of Orkney most invested in kelp production, increased to meet the labor demands of the kelp industry. In fact, it has been demonstrated that population increase was positively associated with per capita kelp production both in Orkney and in the Western Isles (Gray 1957;

Thomson 1983). The price of kelp collapsed in 1830, when the British government repealed tariffs on imported sources of alkali. The end of the kelp boom began a period of economic stagnation in Orkney, as the prior emphasis on the production of kelp was a disincentive to agricultural improvement. Kelp-making was labor intensive and the kelp season fell during the part of the year between planting and harvest when farmers were traditionally underemployed and could otherwise devote time to improvement projects (Thomson 1983, 2008a). The Orkney economy was further limited in the 1830‘s by the nearly simultaneous collapse of the local linen market (Thomson 2008a).

Eventually, Orkney farmers adjusted to the collapse of the kelp market and made agricultural improvements. Inefficient run-rig field systems were abolished and farms were squared off (Thomson 2008a). Common pasture was divided, new fertilization strategies (such as liming) were employed, and field drainage systems were built (Dodgshon 1994; Schrank and

13 Lockhart 1995). Production was also intensified through the laborious reclamation of common pasturage and hill land (Thomson 2008a). However, the kelp boom left a legacy of under-small crofts, which housed laborers during the kelp boom, but were unsustainable as purely agricultural units. The process of reclamation and the addition of new holdings ended by the 1880‘s, after which new holdings stop appearing and small crofts begin to be absorbed into neighboring farms

(Thomson 2008a).

Agricultural improvement also included the addition of new crops into the system of rotation, most notably potatoes and turnips. The production of turnips was essential to the growth of the cattle market, but was previously incompatible with the kelp season. With improvements in steamship transportation, cattle exports increased. This period of expansion continued until the

1880‘s, when the price of cattle fell dramatically. A secondary uptick in kelp production associated with the use of kelp in the production of iodine antiseptics began in the 1850‘s, but declined in the 1880‘s as well (Thomson 1983). Grain prices also fell during this period because of competition from overseas grain markets. However, acreage devoted to grain production remained steady throughout the agricultural downturn because oats were fed to poultry as egg production became increasingly important (Thomson 2008a). This agricultural downturn, in combination with historical patterns of out-migration from the islands, contributed to population decline, which has continued into modern times (Anderson and Morse 1993a, 1993b; Thomson

2008a). The population of the Northern Islands reflects this agricultural boom and bust cycle, as the population of most of the islands reached its height in the mid- to late-nineteenth century

(Table 1-1).

The economic context of the Northern Orkney Islands from 1851 to 1901 represents the end of a period of economic expansion driven by agricultural improvements. The agricultural downturn of the 1880‘s, caused by falling grain and cattle prices was followed by a period of stagnation that lasted through World War I (Thomson 2008b). This period, in which the islands

14 reached their maximum population and then began the long process of population decline, presents an opportunity to study the effects of changing social and economic conditions on household demography, economy, and composition.

Outline

This thesis is a collection of three related, but separate research papers. Chapter 2,

―Households in Northern Orkney, Scotland, 1851-1901,‖ begins with an overview of the comparative study of household form and function. The compound household is defined, described, and integrated into existing systems of household classification. Several case studies are presented to aid the interpretation of the compound household and demonstrate how census listings of household inhabitants relate to the physical structures present at the farmstead occupied by the household. Finally, a larger sample of households is described and comparisons are made to published studies of other preindustrial populations.

Chapter 3, ―The formation and dissolution of extended-family and compound households in Northern Orkney, Scotland, 1851-1901,‖ examines transitions between household types using discrete-time event history models. The influences of several household-level covariates on these transitions, which indicate dynamic processes within households, are described and the dynamics of compound and non-compound households are compared.

Chapter 4, ―Household-level predictors of the presence of life-cycle servants in Northern

Orkney, Scotland, 1851-1901,‖ uses Poisson regression to explore the role of household-level variables in predicting the number of life-cycle servants present in the household. Models of the number of male and female servants are compared to determine whether households hire male or female labor preferentially based on household-level characteristics.

15 Chapter 5 presents an overview of the findings of Chapters 2 through 4 and offers a discussion of the contributions of this work. Future directions for research are also explored.

16

Tables and Figures

Figure 1-1. Landsat-5 satellite image showing the location of the Orkney Islands.

17

Figure 1-2. Landsat-5 satellite image of the six islands included in the study area.

18

Figure 1-3. Histogram of reported ages, 1901 census records from all six study islands. Note the

age heaping, N=4910.

.04

.03

.02

Density

.01 0 0 20 40 60 80 100 Age in years of individual

Table 1-1. Population of the Northern Orkney Islands, 1851-1901. Data taken from census returns, maximum populations noted with asterisks.

North Papa Year Eday Ronaldsay Westray Faray Sanday Westray Total 1851 946 526 371 69 2004 2083 5999 1861 897 531 392* 82 2143* 2151 6196* 1871 994* 539 370 83* 2048 2074 6108 1881 730 547 345 72 2071 2195* 5960 1891 646 502* 287 58 1929 1963 5385 1901 596 441 295 47 1723 1808 4910

19

References

Anderson, M. and D.J. Morse. 1993a. "High fertility, high emigration, low nuptiality: Adjustment processes in Scotland's demographic experience, 1861–1914, Part I." Population Studies 47(1):5-25.

—. 1993b. "High fertility, high emigration, low nuptiality: Adjustment processes in Scotland's demographic experience, 1861–1914, Part II." Population Studies 47(2):319-343.

Ashmore, W. and R.R. Wilk. 1988. ―Household and community in the Mesoamerican past.‖ Pp. 1-27 in Household and Community in the Mesoamerican Past, edited by W. Ashmore and R.R. Wilk. Albuquerque: University of New Mexico Press.

Bonneuil, N. 1997. Transformation of the French Demographic Landscape, 1806-1906. Oxford: Clarendon.

Brennan, E.R. 1979. Kinship, Demographic, Social and Geographic Characteristics of Mate Choice in a Small Human Population. Dissertation, Anthropology, Pennsylvania State University.

Bumpass L.L. and R.K. Raley. 1995. ―Redefining single-parent families: Cohabitation and changing family reality.‖ Demography 32(1):97-109.

Chayanov, A.V. 1931. ―The socio-economic nature of peasant farm economy.‖ Pp. 144-147 in A Systematic Source Book in Rural Sociology, Volume II, edited by P.A. Sorokin, C.C. Zimmerman, and C.J. Galpin. Minneapolis: University of Minnesota Press.

—. 1986. The Theory of Peasant Economy. Madison: University of Wisconsin Press.

Dodgshon, R.A. 1994. "Budgeting for survival: Nutrient flow and traditional Highland farming." Pp. 83-93 in The History of Soils and Field Systems, edited by S. Foster and T.C. Smout. Aberdeen: Scottish Cultural Press.

Fenton, A. 1997. The Northern Isles: Orkney and Shetland. East Linton: Tuckwell Press.

Firth, R. 1966. Housekeeping among Malay Peasants. 2nd Edition. London: Athlone Press.

Goody, J. 1976. Production and Reproduction: A Comparative Study of the Domestic Domain. Cambridge: Cambridge University Press.

Gray, M. 1957. The Highland Economy, 1750-1850. Edinburgh: Oliver and Boyd.

Gray, R.F. 1964. ―Introduction.‖ Pp. 1-33 in The Family Estate in Africa: Studies in the Role of Property in Family Structure and Lineage Continuity, edited by R.F Gray and P.H. Gulliver. London: Routledge and Kegan Paul.

20 Hammel, E.A. 1984. "On the *** of studying household form and function." Pp. 29-43 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

Hirth, K.G. 1993. ―The household as an analytical unit: Problems in method and theory.‖ Pp. 21- 36 in Prehispanic Domestic Units in Western Mesoamerica: Studies of the Household, Compound, and Residence, edited by R.S. Santley and K.G. Hirth. Boca Raton: CRC Press.

—. 2009. ―Craft production, household diversification, and domestic economy in Prehispanic Mesoamerica.‖ Archaeological Papers of the American Anthropological Association 19(1):13-32.

Itzigsohn, J. 1995. ―Migrant remittances, labor markets, and household strategies: A comparative analysis of low-income household strategies in the Caribbean Basin.‖ Social Forces 74(2):633-655.

Jennings, J.A., and Wood, J.W. 2008. ―Multi-census methods for anthropological demography: Applications to the Northern Islands of Orkney, Scotland. Presented at the annual meeting of the Human Biology Association, Columbus, OH.

Jonker, M.A. and A.W. van der Vaart. 2007. "Correcting missing-data bias in historical demography." Population Studies 61(1):99-113.

Kasakoff, A.B. and J.W. Adams. 1995. "The effect of migration on ages at vital events: A critique of family reconstitution in historical demography." European Journal of Population 11(3):199-242.

Laslett, P. 1969. ―The household in England over three centuries.‖ Population Studies. 23(2):199-223.

—. 1977. "Introduction." Pp. 23-89 in Household and Family in Past Time, edited by P. Laslett and R. Wall. Cambridge: Cambridge Univ. Press.

—. 1984. "The family as a knot of individual interests." Pp. 353-382 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

Lee, R.D. 1993. "Inverse projection and demographic fluctuations: A critical assessment of new methods." Pp. 7-28 in Old and New Methods in Historical Demography, edited by D.S. Reher and R. Schofield. Oxford: Clarendon Press.

Levin, I. 2004. ―Living apart together: A new family form.‖ Current Sociology 52(2):223-240.

Murtha, T.M., P.L. Johnson, J.W. Wood, S.A. Matthews, C.S. Sparks, and J.A. Jennings. 2008. "Historical demography, oral history, settlement archaeology, and landscape ecology: The North Orkney Population History Project." Presented at the annual meeting of the European Population Conference, Barcelona, Spain.

21 Netting, R.M. 1993. Smallholders, Householders: Farm Families and the Ecology of Intensive, Sustainable Agriculture. Stanford: Stanford University Press.

Netting, R.M.C., R.R. Wilk, and E.J. Arnould. 1984. "Introduction." Pp. xiii-xxxviii in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R.R. Wilk, and E.J. Arnould. Berkeley: University of California Press.

Oeppen, J.E. 1993. "Generalized inverse projection." Pp. 29-39 in Old and new methods in historical demography, edited by D.S. Reher and R. Schofield. Oxford: Clarendon Press.

Ruggles, S. 1992. "Migration, marriage, and mortality: Correcting sources of bias in English family reconstitutions." Population Studies 46:507-522.

Sana, M. and D.S. Massey. 2005. ―Household composition, family migration, and community context: Migrant remittances in four countries.‖ Social Science Quarterly 86(2): 509-528.

Santley, R.S. and K.G. Hirth. 1993. ―Household studies in Western Mesoamerica.‖ Pp. 3-17 in Prehispanic Domestic Units in Western Mesoamerica: Studies of the Household, Compound, and Residence, edited by R.S. Santley and K.G. Hirth. Boca Raton: CRC Press.

Schrank, G. and D.G. Lockhart. 1995. An Orkney Estate: Improvements at Graemeshall, 1827- 1888. East Linton: Tuckwell Press.

Sieder, R. and M. Mitterauer. 1983. "The reconstruction of the family life course: theoretical problems and empirical results." Pp. 309-346 in Family Forms in Historic Europe, edited by R. Wall, J. Robin, and P. Laslett. Cambridge: Cambridge University Press.

Smock, P.J. 2000. ―Cohabitation in the United States: An appraisal of research themes, findings, and implications.‖ Annual Review of Sociology 26:1-20

Sparks, C.S. 2007. Households, Land, and Labor: Population Dynamics in the Northern Orkney Islands, Scotland, 1851 to 2003. Ph.D. Dissertation, Anthropology and Population Research Institute, Pennsylvania State University.

Thomson, W.P.L. 1983. Kelp-making in Orkney. Stromness: The Orkney Press.

—. 2008a. The New History of Orkney. Edinburgh: Birlinn Ltd.

—. 2008b. Orkney Land and People. Kirkwall: The Orcadian Limited (Kirkwall Press).

Wilk, R.R. 1991. Household Ecology: Economic Change and Domestic Life Among the Kekchi Maya in Belize. Tucson: University of Arizona Press.

Wilk, R.R. and R.M.C. Netting. 1984. "Households: Changing Forms and Functions." Pp. 1-28 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R.R. Wilk, and E.J. Arnould. Berkeley: University of California Press.

Winfield, F.E. 1985. Commuter Marriage: Living Together, Apart. New York: Basic Books.

22 Wood, J.W. Forthcoming. Population, Food, and Traditional Farming: A New Synthesis.

Wrigley, E.A. 1994. "The effect of migration on the estimation of marriage age in family reconstitution studies." Population Studies 48(1):81-97.

Wrigley, E.A., R.S. Schofield, R.S. Davies, J.E. Oeppen, and C. Wilson. 1997. English Population History from Family Reconstitution. Cambridge: Cambridge University Press.

Yanagisako, S.J. 1979. "Family and household: The analysis of domestic groups." Annual Review of Anthropology 8(1):161-205.

Chapter 2

Households in Northern Orkney, 1851-1901

Introduction

The study of household form and function in the past was revolutionized by the digitization and comparison of census (or census-like) listings of communities. These studies suggested that in Northwest Europe, notions of the ubiquity of large extended-family households were mostly a function of nostalgia and long held assumptions about family life in the past and could not be supported by historical evidence (Hajnal 1965, 1982; Laslett 2000; Laslett et al.

1983; Laslett and Wall 1974). However, later studies have shown that there is more regional variation in household size and composition than the original descriptions of the nuclear family dominated household systems of preindustrial Europe would have suggested (Berkner 1972; Czap

1982; Engelen and Wolf 2005; Hammel 1972; Kertzer 1984; Szoltysek 2008).

This study describes households in Northern Orkney, Scotland, in the period between

1851 and 1901. The treatment of households in this chapter is purely descriptive, whereas

Chapters 2 and 3 present formal analyses of household dynamics and economics. Households are examined using multiple data sources, including individual-level census returns, historical estate and ordinance survey maps, and archaeological surveys of farmsteads. Together, these data sources provide evidence for a household form, which I will call the compound household, that would otherwise go undetected if only census returns were used to determine household membership. This household form is characterized by the presence of multiple, but economically interdependent dwelling units. Put another way, compound households consist of groups of related people who reside in adjacent dwellings but share a common set of essential farming structures. When the prevalence of compound households in this population is compared to that

24 in other areas of Northwest Europe, it appears that Orkney is characterized by higher than usual levels of household complexity. In light of this evidence and other regional studies that indicate local variation in household forms, broad generalizations about the frequency of the nuclear family household in Northwest Europe should be reconsidered.

Background

Setting and Data Sources

The Orkney Islands are an archipelago off the coast of northern Scotland. The six islands included in the study area are Westray, Sanday, Papa Westray, Eday, North Ronaldsay, and

Faray. These islands were selected as part of an ongoing multidisciplinary study of population and family history, settlement, and land use known as the North Orkney Population History

Project (Jennings 2010; Jennings et al. 2009; Murtha et al. 2008; Sparks 2007, 2009).

The northern Orkney Islands were, and have remained, rural in character. Traditionally, agricultural production relied on grains, mainly black oats and bere (a landrace of barley), and livestock, including sheep, cattle, chickens, and pigs. These activities supported household subsistence and paid rents (sometimes in kind). Individual farmsteads, or groupings of dwellings and agricultural buildings associated with gardens and fields, are dispersed over the islands.

There are a few small villages that are more densely settled, but the villages are still primarily agricultural in nature. Farmsteads have names that appear in historical records and persist for many generations, even while the inhabitants change over time (Fenton 1997; Palsson and

Edwards 1981; Thomson 2008a, 2008b). Cadastral (estate) maps from the decades before the study period, such as the 1830s and 1840s, feature farmsteads with the same names and locations as in later historical and modern maps. Orkney historians speculate that most of the farm and

25 house names became fixed by the 1840s to accommodate record keeping (such as census enumerations and valuation rolls) and mail delivery (Thomson 2008a). The population of the islands reached it height in the mid- to late-1800s, after which out-migration, both to mainland

Scotland and overseas, began the depopulation that continues to this day. Depopulation and the widespread use of stone as a building material have facilitated the survival of archaeological remains of farmsteads. The remains allow for the study of the settlement history of the islands and provide information about the structures present at farmsteads that appear in historical records.

Farmstead names are listed in census returns, along with information about the inhabitants of the farmstead and their relationship to the household head. With this information, household size, composition, and type can be tracked over time. Valuation rolls, which are records of property values against which taxes were assessed, provide a good proxy measure of landholding quantity and quality, and are linked to census returns using farm names (see

Appendix A for details about how this assessment is made). These written sources, in combination with information from estate maps (where available) and historical Ordinance

Survey of Great Britain (OSGB) maps, as well as historical archaeological surveys of abandoned farmsteads conducted by NOPH personnel provide information about the composition of households, the buildings associated with the farmsteads they occupied, and how these arrangements have changed over time.6

6 Archaeological surveys were conducted under the supervision of Dr. Timothy Murtha, a member of the NOPH project. Specific farmsteads were selected based on the team‘s ability to access the site. This usually meant that the farmstead was no longer occupied and that permission was granted by the current landowner.

26 Household Types and Categories

Scholars associated with the Cambridge Group for the History of Population and Social

Structure developed an approach to the study of family history using censuses or census-like listings of communities that was both widely adopted and criticized (Berkner 1975; Hammel

1984; Laslett et al. 1983; Laslett and Wall 1974; Yanagisako 1979). These studies were conducted to provide information about the domestic context in which past people lived, worked, and made economic and demographic decisions. In addition, the suggestion of a common set of terms and a standardized system of household classification, although perhaps not ideal or universally implementable, allowed for cross-cultural comparisons and the study of change over time within populations. The comparative approach remains popular, although the focus has shifted from comparing household form and function to investigating themes such as adaptation to economic pressure or the timing of demographic events, such as marriage, best typified by the work of the EurAsia project (Bengtsson et al. 2004; Bengtsson and Mineau 2008; Bengtsson and

Saito 2003; Engelen and Wolf 2005; Neven and Capron 2000).

Among the most important findings of the Cambridge Group was the discovery that preindustrial Europe was characterized by nuclear, simple-family households rather than the large, extended families that were previously associated with this period. However, subsequent work has demonstrated that there existed significant regional variation in family forms in

Northwest Europe that was overlooked by earlier studies (Anderson 1971; Engelen and Wolf

2005; Janssens 2002; Kertzer and Barbagli 2002; Laslett et al. 1983; Reay 1996; Ruggles 2009).

As this more recent work has shown, our understanding and interpretation of households requires reconsideration as new information becomes available. This study uses information about households in Orkney that comes from a variety of sources typically not available to historical demographers, most importantly historical maps and archaeological information. These sources

27 place census listings of households in the context of the physical dwelling and farming buildings essential to both the spatial and economic organization of domestic groups and their farmsteads.

The use of multiple data sources in the analysis and interpretation of household structure and function addresses some of the issues raised by critics of the Cambridge Group‘s approach to the study of English communities (Berkner 1975).

For example, data that supplement census listings can improve our understanding of the meaning of boundaries between groupings of individuals in these documents, thus avoiding some of the interpretative ambiguities that are common to studies that rely solely upon written lists of group members (Berkner 1975; Hammel 1984). Despite the advantages of this approach, this study cannot resolve the recurring issue of whether or not the household is a reliable unit for comparative research or how it should be defined and measured in general (Hammel 1984).

Rather, it demonstrates how the use of additional data sources enhances the interpretation of census listings and reveals a type of living arrangement that would otherwise be overlooked.

Household and Farming Unit

While there are many definitions of the household, they usually include at least one or two of the following criteria: location, shared activity, and kinship (Laslett 1977). In Orkney, location and kinship can be determined through census enumerations and vital records. The study of shared activity in the past is more difficult, however. Analogies can be drawn from studies of other traditional agrarian societies, in which the household is also the major farming unit (Netting

1993). Evidence from Orkney is consistent with this generalization. Members of the household provided labor inputs for agricultural production and domestic work, and they could expect to be provided food, shelter, and other necessities from the pooled output of the farming unit. The

28 isomorphism between the household and the day-to-day farming unit is central to the definition and description of the household used in this paper.

Households in Orkney: Simple, Extended, and Compound Types

For the purposes of this study, household types are defined following the classification system devised by Hammel and Laslett (1974), with one notable exception. In this system, simple households consist of a married couple, or a surviving member of that couple, and any of their co-resident children.7 Extended households, in contrast, feature one or more related individuals in addition to the simple household. This definition merges what other family history researchers would call joint and stem households. A third household type, not included in

Hammel and Laslett‘s original classification system, is found in Orkney. This type is identifiable because the surviving archaeological remains in Orkney provide evidence for a household type that commonly-used historical records, such as census listings, cannot adequately describe.

While demographic data sources suggest that extended households are less common than simple households, the physical remains of household-based farmsteads reveal that many units listed separately in the census were actually single households in the economic sense described above

(see Hammel 2005, Wilk 1991, and Netting 1993 for other descriptions of the economic functions of household units). These compound households, often linked by brothers, are adjacent or even structurally joined, and share a common set of essential farming structures such as barns, byres, grain kilns, and stables. Given what has been learned about Orcadian economic and social systems through this interdisciplinary project, the investigators are confident that the component units of such compound households were not independent, but rather worked their holdings

7 These definitions refer to related individuals within the household. The presence of servants, lodgers, boarders, and visitors is not inconsistent with the simple household as defined here. Thus, a household that contains a married couple, their children, a servant, and a lodger is still classified as a simple household.

29 cooperatively and shared the products of their labor, even if employed in outside wage labor, which was usually part-time or seasonal. By this criterion, the sharing of a single set of farming structures, over the period 1851-1901, compound households comprise 30-44 percent of households in the study area, whereas extended households represent between 17 percent and 22 percent of households (Table 2-1).

In census listings from Orkney, several household heads and their families are sometimes associated with the same named farmstead location. Traditionally, a historical demographer would interpret such listings as evidence of two or more independent households that lived nearby, much in the same way that one would consider the residents of an apartment building as independent households occupying a common structure (see Laslett 1977, Hammel and Laslett

1974, and Berkner 1975 for detailed discussions of how historical demographers normally interpret census-like listings). However, archaeological investigation and detailed mapping suggests that in Orkney this interpretation is incorrect. Further examination of these households has shown that when multiple heads are listed at the same location, multiple dwellings are also found, but these dwellings (usually very close or attached) share a common set of features, such as barn, byre, stackyard, kailyard (house garden), and muckyard (manure midden) needed for a single farming unit. Historical maps indicate that these structures existed contemporaneously with the census listings of multiple heads. Whatever else they may do, including part-time and seasonal work on other farms or as craftsmen, crofters are farmers on their own holdings and everyone living at the croft helps with farm production if physically able. Therefore, these compound households should be considered single (although complex) households whose members pool resources as one united farming unit, rather than as two or more independent households that happen to occupy the same site.

This type of compound household is not limited to Orkney. Compound households, in which dwelling units are contained within an outer wall and whose members share production

30 activities and commonly pool resources, are found throughout West Africa, usually in association with polygynous households (Goody 1958; Goody 1972). In his study of a Maya group in Belize,

Wilk describes ―household clusters‖, or households composed of more than one dwelling unit that share economic activities (1984; 1991). Arrangements similar to those described below in case studies from Orkney, where a pair of siblings and their families cooperatively work a single farm while living in separate dwellings, have also been documented in 19th century France (Segalen

1983, 1984). Russian peasant households have similar features, as a complex household may have multiple izby (dwelling units) but a single set of farming structures, such as barns (Czap

1983). In addition, Berkner (1975) disagrees with some of Laslett‘s definition of households, arguing that in Austria, retired parents living in separate quarters or a cottage on the same premises as their children should be considered a single household rather than two independent households. He further suggests that the interpretation of household divisions and boundaries is a difficult task that should be based on knowledge of local contexts (Berkner 1975). Fortunately, archaeological and documentary information from Orkney provide information about local context that aids in the identification of compound households and, in many cases, the reconstruction of household histories and kinship links within the household as well. This paper presents examples of the kind of household reconstruction that is possible when multiple data sources are used.

Orkney Farmsteads: Buildings and Furnishings

Orkney farmsteads are characterized by long rows or parallel rows of dwelling and farm buildings. Early farms (built well before the study period) were influenced by Norse building customs (Dunbar 1966). Longhouses, sometimes called ―byre-dwellings‖, housed people and livestock under the same roof (Fenton 1997). Early farms also featured central hearths, ventilated

31 though a smoke-hole, which also let in light. By the second half of the nineteenth century, however, farmstead architecture had changed, and the surviving remains of farmsteads visible today reflect these changes. In these ―improved‖ designs, farm buildings were often separated from dwelling buildings by a walkway or close. Central hearths were replaced by gable hearths with chimneys. Dwellings generally consisted of a either a single room, subdivided into two parts, or two rooms, known as the ―but‖ and ―ben‖ ends. The but end housed the kitchen and larger cooking hearth. The ben end was an inner room devoted to living space with a smaller fireplace for heat and light. Often box-beds, or sleeping areas enclosed on three sides by wooden panels, separated the but and ben ends.

Byres, barns, stables, and corn kilns could be structurally joined to dwellings in a single line. Other common farmstead layouts featured two parallel blocks of buildings or blocks of buildings situated at right angles to each other (Fenton and Walker 1981). Both dwellings and farm buildings were constructed primarily of stone, with flagstone roofs sometimes overlain with thatch. Generally, a kailyard, muck pile, and stackyard could be found in close proximity to the farm buildings. In the case of compound households, two or more dwelling units, identified by their distinctive gable hearths and chimneys found at the but and ben ends, shared a single set of farm buildings (barn, byre, stable, etc.) and a single set of outdoor areas (kailyard, stackyard, muck pile, and plots of arable and pasture).

Household Histories using Multiple Data Sources: Case Studies

The following case studies illustrate the kind of household history that can be reconstructed using census records, vital registers, historical maps, and archaeological surveys.

The inhabitants of each farmstead are indentified in individual-level census returns. Their genealogical relationships are reconstructed from the ―relationship to head‖ field in the census

32 form and from civil records of births, deaths, and marriages. Occupations are often listed in census returns. These occupations may represent either full-time work or, more frequently, part- time or seasonal bye-employments that usually provided wages.8 For example, an individual may be listed as a plowman, but plowing was a seasonal activity. In these cases, the listed plowman presumably worked on his own plot, but would also be hired (either by the day or ―piece‖ of work) to plow on larger farms. The occupation of kelp-maker is another example of seasonal work. Kelp was produced by hauling seaweed to shore and burning it to create a marketable product used to produce alkali for the manufacture of soap, glass, and dye (Thomson 1983).

However, this work could only be conducted in the months between planting and harvest. While kelp-making was labor-intensive and presumably a full-time job during the months of the kelp season, kelp could not be produced year-round and individuals listed as kelp-makers probably worked on the farm when kelp work was not available. Thus, while the listed occupations provide some useful information about the activities of the inhabitants of a farmstead, it may be the case that a single person participated in several lines of work in addition to helping with the farm. Indeed, the situation regarding bye-employments in Orkney was probably much like that described by Ian Whyte in a study of Westmorland. He writes, ―it is clear that a number of people listed as, for example, carpenters, tanners, and weavers in the census occupied enough property to qualify for substantial allotments of up to 14 acres and were really husbandmen with part-time trades‖ (2007:102).

Building types are indentified, when possible, from archaeological surveys of the remains of farmsteads. These identified buildings are then labeled on digitized or photographed historical

8 A few occupational categories require additional clarification. An individual listed as the farmer presumably farms for himself. Those who ―assist on the farm‖ are usually younger household members who work under the direction of the farmer. Individuals whose occupation is listed as ―farm servant‖ or ―domestic servant‖ live as members of the household, but work as a servant for another household, usually for wages. This might represent full or part time work, piecework, or day labor. In contrast, individuals whose relationship to head (not simply occupation) is listed as ―servant‖ are live-in life-cycle servants on the farmstead in question, who receive room and board in addition to wages. For a discussion of the differences between day laborers and life cycle servants, see chapter 4.

33 maps, which can often show changes in the number and type of buildings present at a farmstead over time. Modern maps and aerial photographs are included for reference and to give a sense of the state of preservation of farmstead remains. Archaeological survey was essential for indentifying the functions of farm buildings. Dwellings are characterized by fireplaces with chimneys that are situated at the gable ends of the rooms. Dwellings also feature windows, and may have remains of furniture, such as box beds, or domestic equipment, such as stoves and cooking implements. In contrast, byres did not generally have windows. Upright flagstones that functioned as stall dividers can be used to identify byres, as can the presence of muck troughs and oddle holes, which facilitated the removal of animal wastes to the muck yard, where they were collected for manuring. Stables share many features with byres, but are generally smaller, and may have the remains of wooden stalls (often with bite marks left by horses), hay mangers, and sometimes horse harnesses. Barns are easily identified by the presence of two doors. These doors are aligned opposite each other, and could be opened to provide a crosswind when grain was threshed and winnowed. Corn drying kilns are often found attached to barns. Generally, these structures were round, though square examples exist, and could be accessed from inside the barn to add and remove grain and to tend the fire. Corn kilns are thought to be a distinctive feature of the Scottish Islands, and are less common on mainland farms (Naismith 1985). Hen houses often feature roosts or nest boxes, and pig houses are generally small, with a low roof.

Unlike barns and byres, a single farm could have multiple hen and pig houses.

South Hammer, Westray: Combining data sources for new interpretations

South Hammer was a 10-acre croft on the island of Westray. The inhabitants of South

Hammer in each census year (1851-1901) are shown in Table 2-2. Figure 2-1 illustrates the genealogical relationships among these individuals in the 50-year interval. At its height, the

34 compound consisted of three dwellings, two of which, along with a barn and a grain-drying kiln, were arranged in a single linear block (Figure 2-2). A flagstone pathway (close) separates this block from the byre, stable, and pig houses. A cluster of buildings associated with a smithy is also separated from the main block. A third dwelling was constructed after (and somewhat further from) the other buildings at the northwest corner of the kailyard (house garden enclosure).

In 1851, South Hammer is occupied by a farmer, Charles Paterson (1), his wife Ann (2), and their children.9 At this time, South Hammer is a simple family household. In 1861, their second son, Stewart (4), has become a blacksmith. The smithy buildings may date to near this time. By 1871, changes have come to the croft. It is now a compound household, headed by

Charles (1) and his sons Stewart (4) and William (8), and their wives and children. Charles is still listed as the farmer of South Hammer, while Stewart and William are blacksmiths. Given this change in family structure, the additional dwelling to the north of the main house probably dates to this time. The easternmost dwelling may also represent an addition to accommodate new household units. If an estate map that pre-dates 1871 can be found for this region of Westray, these speculations may be verified (see the use of estate maps in the description of Sangar below).

In 1881, the household retains much of the same configuration, but now William (8) is listed as a fisherman. The Ordinance Survey map from 1882, in combination with archaeological surveys of the remaining buildings demonstrate that each family unit likely occupied a separate dwelling, while sharing a single set of outbuildings, including the barn, kiln, byre, stable, pig houses, and smithy buildings (Figure 2-2). By the census of 1891, Charles (1) and Ann (2) have died, and Robert (3), their eldest son, has returned to South Hammer with his family. Now, the heads of the three units of the compound household are all brothers and Robert (3) is now identified as the farmer. In 1901, Stewart (4) and his family have left South Hammer, and Janet

(9), now a widow, has returned to her childhood home with her children. Oral histories indicate

9 Numbers associated with individuals correspond to numbers in the tables of inhabitants and genealogies.

35 that Janet (9) and her children occupied the northern dwelling, Robert (3) and his family occupied the western dwelling in the main block, and William (8) and his family occupied the eastern dwelling.10

South Hammer represents the most detailed reconstruction of a farmstead and its history that is possible with current data sources. One of the dwellings was occupied until 1989, and the buildings remain remarkably well preserved and are now under the care of the Westray Buildings

Preservation Trust, which had restored and preserved this traditional farmstead (see Figure 2-3 for modern photos of the croft complex). This has allowed for thorough archaeological study combined with information gathered from historical records and the neighbors and relatives of the last living inhabitant, who died in 1993. Indeed, South Hammer was the starting point for much of our work with compound households, as it first revealed that households listed in the censuses with separate ―heads‖ (but with a shared farm name) could be occupied by a single economically- interdependent group of relatives.

Sangar, Westray: Repurposing buildings

The farmstead of Sangar provides an interesting example of how households repurposed the physical structures of their farmstead to accommodate changes in household size and composition (Table 2-3, Figure 2-4). Sangar was a croft of 15 acres, consisting of two blocks of buildings perpendicular to each other. One block contains a dwelling, hen house, and pig house.

The other block features a dwelling, barn, kiln, and byre. The order of construction of the blocks is not known, but the farm dates to at least 1848, when both blocks were included in a map of the

10 In Tables 2-2 through 2-9, the order of the constituent parts of compound households (when present) may change. For example, the inhabitants of South Hammer 1 in 1871 are listed under South Hammer 2 in 1881. It is unclear whether this represents a change in actual dwelling units occupied, or if dwelling units remained relatively constant and the enumerator simply recorded the units in a different order. Given the time and effort required to move house, I speculate that the latter is true in most instances.

36 Rapness estate (Figure 2-5). Although now abandoned, this farmstead was occupied until the modern period, and some of its former inhabitants are still living. This relatively recent occupation has contributed to the excellent preservation of the buildings, which have been subject to archaeological survey.

In 1851, Thomas Drever (1) and his family lived at Sangar. Thomas was a farmer and head of a simple-family household. The household remained largely unchanged until 1871, when

Thomas‘s youngest son, William (6) is listed as a fisherman. In 1881, Sanger became a compound household, with Thomas (1) heading one unit, and William (6), now married with children, heading the other. Thomas is now the farmer of Sanger, while his son is a sailor in the merchant service (one of the few available fulltime employments). In 1891, the household arrangements are largely the same, but now William is also listed as a farmer. He has presumably returned from sailing to help work the farm, as his father is now 83 years old. His mother had died shortly after the 1891 census enumeration, and his father would pass away in 1897. Thus, by

1901, the household has returned to a simple-family household.

This household history is a good example of what is known as the stem-family household system. In this example, the youngest son is the heir to the farmstead and remains at home while his siblings leave. When he marries, he and his wife and children co-reside with his parents until their death.

Unlike at South Hammer, it is not possible to determine which nuclear families resided in which dwellings at Sangar. Between 1871 and 1881, one of the blocks may have been repurposed to accommodate the formation of a compound household, as both blocks were present since 1848. In 1881 and 1891, Sangar was a compound household, and the two dwelling units shared a single set of farm buildings. Later (the exact timing is unclear, but presumably post-

1891), the dwelling in the western block was repurposed (perhaps for the second time) into a

37 byre, and the fireplaces were closed off and stalls and muck troughs were added to accommodate cattle (Figure 2-5, see also Figure 2-6 for a modern photograph).

Quoynabreckan, Westray: Stability over time

Quoynabreckan (a croft of 5 acres expanded to 15 by 1861), is an example of a non- compound household (Table 2-4 and Figure 2-7). Quoynabreckan was architecturally simple, consisting of only a single block of buildings. From north to south, the complex contained a kiln, barn, dwelling, and byre/stable, slightly offset from the rest of the block (Figure 2-8, 2-9).

At the beginning of observation, a farmer named John Rendall (1) and his family occupied Quoynabreckan. In 1861, the family changed from a simple-family household to an extended family household when his grandson, also John (11), came to live at the farmstead (his parents presumably lived elsewhere). By 1871, John (1) has died and his son John (7), now married with children, has inherited the farm. His mother Margaret (2) is listed as shareholder of the farm, a sign of a possible retirement agreement between mother and son. John (7) has hired a farm servant. This hiring probably represents a reaction to a shortage of household labor, as John and his wife must support both his aging mother and their infant son.

After Margaret (2) died, the household changed from an extended-family household to a simple family household. This transition in household form occurred because of the death of one household member. This is an important characteristic of the dynamics of extended households, especially when compared to compound households (see Chapter 3 for formal analysis). This simple-family arrangement, headed by John (7), persists until the end of observation in 1901. In this example, we see another instance of a stem-family, as one son inherited the farm upon the death of his father. The young family co-resided with the widowed mother until the time of her death. However, because there was only one dwelling present at Quoynabreckan, this household

38 is classified as extended in the years that the grandson and the widowed mother resided (1861 and

1871), rather than compound.

Ha’Ouse, Westray: An unusual arrangement of buildings

Although it is a croft of modest size (4 acres later expanded to 7), Ha‘Ouse features an uncommon number and diversity of buildings, some of which are difficult to interpret. The

Rousay family (Table 2-5 and Figure 2-10) occupied this architecturally interesting farmstead.

On the southern end of the complex are two dwellings, identified by their fireplaces and widows

(Figure 2-11, 2-12). A block of buildings runs perpendicular to the dwellings and includes another possible dwelling adjacent to the other two, a stable (with flagstone stall partitions), and a byre. Another block of buildings extends westward perpendicular to the byre. From east to west, there are two rooms with fireplaces, a small building with windows, a barn, and a kiln. Several small buildings, whose functions are difficult to interpret, are attached to the south wall of the barn. Those with fireplaces and widows may represent workshop areas (e.g. weaver‘s sheds), as many of the inhabitants of Ha‘Ouse were involved in woolen production or dressmaking.

Unfortunately, pre-1871 maps of this farmstead are not available at this time, so the timing of the construction of segments of the complex cannot be dated well.

Ha‘Ouse was home to a compound household throughout the study period. From 1851 to

1871, the compound household was headed by John Rousay Sr. (1) and John Rousay Jr. (3).

After the death of John Sr., Margaret (4), who had been keeping house for her father, assumed headship of the other unit of the compound household. John was listed as the farmer, and

Margaret was a spinner. Unlike the case of South Hammer, it is not possible to determine which portion of the compound household occupied which dwelling. Interestingly, Margaret‘s status

39 switches from ―head‖ to ―lodger‖ and back between 1881 and 1901. The implications of this change in status for Margaret‘s role in the household are unclear.

Stanegeo, Westray: Adding a new dwelling

Consecutive census observations of Stanegeo (18 acres) begin in 1881, when farmer

David Rendall (1) and his family resided there (Table 2-6, Figure 2-13). Stanegeo originally consisted of a main block of buildings including, from west to east, a byre, dwelling, barn, and kiln (Figure 2-14, 2-15). A small outbuilding (function unclear) was situated south of the main block and is labeled in Figure 2-14.

In 1901, the household changed to a compound household, headed by David (1) and his second son, James (5). David was listed as the farmer, and resided in a dwelling unit with his wife, unmarried children, and illegitimate grandson. James was listed as a fisherman and kelpmaker (a seasonal occupation) and resided with his wife and children. In addition, a second dwelling was constructed east of the main block at approximately the same time (Figure 2-14).

The construction of the second dwelling at Stanegeo probably predated the 1901 enumeration, as it is featured on the 1901 OSGB map of the farmstead, and the family had presumably changed forms around the time of James‘s marriage. The concurrent timing of the construction of the new dwelling and the transition to a compound household form increases confidence that changes in household form, as observed in census records, correspond to changes in dwelling arrangements in the physical farmstead.

40 Nether Brough, Westray: When relationships within a household remain unclear

Stewart Rendall (1) and his descendants live at Nether Brough (2.75 acres) from 1851 to

1901 (Table 2-7 and Figure 2-16). However, in 1851 and 1861, they shared the farmstead with the Drever family, who moved to Nether Banks by the 1871 census. No close genealogical relationship has yet been found between these two families. However, Rendall and Drever are extremely common surnames on Westray. In fact, 22 percent of the population of Westray in

1861 census had either Rendall or Drever as a surname. This finding contrasts with other examples of compound households that are headed by close relatives, such as a father and son, or a set of siblings, so its implications for the definition of compound households should be considered carefully.

For instance, should the inhabitants of Nether Brough be considered a compound household or two independent households in 1851 and 1861? There are several plausible interpretations of this evidence. If one employs the definition of a compound household that uses the criterion of multiple dwellings that share farm buildings, then this household should be considered compound. Multiple dwellings can be indentified at Nether Brough, which consists of two clusters of buildings (Figure 2-17, 2-18, 2-19). The south cluster consists of two parallel blocks. The southern block contains three buildings, a possible workshop (although the exact function is not clear from the archaeology—it features a window and some boxes set into the walls), a byre (identified by a flagstone stall partition), and a possible dwelling or workshop (with a stove, widow, skylights, and access to an attic). The north block contains two dwellings. The second cluster of buildings is at the north end of the farmstead and it consists of only one building block. This block contains a kiln, barn, and threshing room. Thus, the evidence from the physical structures present at Nether Brough support the interpretation of the inhabitants of this

41 farmstead as a compound household. In fact, in 1871 and 1881 the farmstead becomes compound in the classic sense, as relatives of the Rendalls have replaced the Drevers.

However, it is also possible that the two families may have lived in very close quarters, but largely independently. In 1851, the head of the Drever family is listed as a fisherman, while

Stewart Rendall (1) is listed as a weaver and farmer. Thus, it is possible that these two groups of people made independent livings, and should not be considered a compound household in the economic sense. However, we cannot say whether the ―girl‖ employed by Stewart is actually

Isabella Drever, who is characterized as a farm servant in the same census year (Table 2-7). In addition, in 1861, James Drever is listed as a tailor. While he may have carried out his trade independent of the adjacent Rendall family, it seems unlikely given that the Rendalls have a tradition of working as weavers (Table 2-7). Despite these circumstantial lines of evidence, this example illustrates how caution should be taken when describing households, especially complex or compound households.

Moa, Westray: Interpretation without surveys

The farmstead of Moa provides an interesting contrast to the examples given above. Moa was a significantly larger farmstead of 60 acres, consisting of three blocks of buildings (Figure 2-

21). Two parallel blocks are found in the southern end of the farmstead, while a single block runs perpendicular to the other two (Figure 2-21, 2-22). Surveys of Moa have not been conducted, so it is difficult to ascertain the functions of the various buildings. Moa was apparently home to a compound household at each census year with the exception of 1871 (Table 2-8, Figure 2-21).

Given this relative stability in household form over time, the number of individuals present at the farmstead (sometimes as many as 13), and the consistency of buildings present from 1848 to

1901, it is probable that more than one dwelling was present at the farmstead. If this farmstead

42 follows the pattern of compound households observed in other examples, it is also likely that these dwelling units shared a common set of farming buildings, although possibly larger, more numerous, or more elaborate than is typically found with smaller holdings. While this cannot be tested directly at this time, the size and layout of the farmstead, along with the larger than average landholding (60 acres), is consistent with this speculation.

Moa was originally headed by Robert Drever (1) and two of his sons. After the father‘s death, the sons continued to head the component units of the compound household, with the exception of 1871, where the part of the household headed by Robert Jr. (5) is missing. Perhaps some of the outbuildings were associated with carpentry, as several occupants of Moa were listed as joiners, carpenters, and boat builders (Table 2-8). In addition, Fae Quoy tae Castle (Westray

Buildings Preservation Trust 2002) notes that Robert Drever also worked as an undertaker in addition to his activities as a farmer and joiner, which makes sense since the primary duty of an undertaker was to make coffins.

Discussion of Case Studies

From these case studies, it is clear that when multiple heads are listed in the census for a single farmstead, multiple dwellings, but only one set of farm buildings, are also present. Indeed, in some instances, the addition of a new dwelling corresponds to the transition from a non- compound to compound census listing format. This finding bolsters the interpretation of these units as economically interdependent single households. However, in some instances, it appears that the component units of a compound household are not closely related. This finding is contrary to kinship-based definitions of households, and care should be taken in the interpretation of the economic relationships among such coresident, but seemingly unrelated, units. However, it is important to note that complete, comprehensive genealogies have not yet been constructed, so

43 it is not certain that, for example, the Rendalls and Drevers of Nether Brough are not closely related.

In addition, these case studies provide evidence for the existence of stem-families in northern Orkney. These families are easy to identify when a single household is tracked longitudinally, but could be missed in cross-sectional data, as few census observations would capture these households while they were in three-generation form. While case studies, such as the ones presented here, offer insight into how households, and the physical structures they inhabited, change over time and between generations, one cannot characterize a larger study area using only a few detailed examples. A descriptive overview of all the households in the study area for which identifying information is available will better describe the variation in household form over the whole of Northern Orkney from 1851-1901.

Households in North Orkney: A Descriptive Overview

This part of the study describes households in general. Changes in patterns of household size and composition are tracked over time. First, kin relationships within households are examined using census listings, which specify each individual‘s relationship to the head.

Although detailed genealogies have not yet been reconstructed for the majority of households, the census-based approach captures some of the kin relationships within households at single points in time, defined by the enumerations. Age- and sex-specific trends in these kin relationships, as well as headship rates, are examined. Household types (simple, extended, and compound) are compared in terms of size, valuation, and membership over the life course. A detailed breakdown of households following the Hammel-Laslett classification system follows. Compound households and their effects on proportions of household types observed in the population are considered. Finally, cross-cultural comparisons are drawn using published data.

44 The sample used in this portion of the study is drawn from all six islands of North

Orkney. Households whose names could be successfully located on area maps were selected into the sample. This was done to ensure that the constituent units of compound households could be accurately identified and grouped together. When named households cannot be located spatially, it is difficult to ascertain whether the dwelling units actually share a common compound and set of farm buildings. This selection may introduce bias into the sample, as the households that could be located spatially may differ from households that could not. For example, the unlocatable households may represent temporary housing or a group of highly mobile people. Alternately, these households may have occupied small crofts whose existence was never mentioned in written sources or ―forgotten‖ in oral histories.

Living with Kin

It is possible to describe a household in terms of the kin relationships present within the residential group. In the above case studies, this was accomplished through the reconstruction of genealogies. However, comprehensive genealogies of the study‘s majority of inhabitants have not yet been constructed. In the absence of complete genealogical reconstructions, it is possible to characterize households based on the members‘ relationships to the head as listed in the census returns. The most common co-resident extended kin, (those related to the head, but excluding the head and the head‘s spouse or children) were grandchildren (2.9-4.9 percent of the sample), followed by sisters (1.6-2.6 percent of the sample) and mothers (1.3-2 percent of the sample).

These figures suggest that only a small percent of the total population were living with extended kin at any census enumeration (Table 2-9). However, if we examine the percentage of households that contain extended kin, 13-17 percent of households include grandchildren, 10-13 percent of households include widowed mothers, and 10-14 percent include adult sisters (Table 2-

45 10). Further, these figures may underestimate the total time that individuals live with extended kin, as they only consider cross-sectional measurements of household composition at ten-year intervals. In addition, only the constituent ―segments‖ of compound households are considered in

Tables 2-19 and 2-20, as the current state of data linkage only allows kin reckoning to the head of each part of the compound household, rather than within the entire compound household. This also contributes to the underestimation of the percent of households that include extended kin.

Even with these limitations in mind, it is possible to get a general sense of how kin relationships change over individual life courses. Again, using individual-level census returns, broken down by the sex and age of individuals, some sex- and age-specific trends in kin relationships emerge. For males (Figures 2-23, 2-25, 2-27, 2-29, 2-31, 2-33), the percent living as children of the head decreases, while the percent of heads increases as they age and assume headship. There is a peak in the percentage living as non-kin in the late teens and early-twenties, probably a result of the common practice of life-cycle service.11 Finally, there is an increase in the percentage living as relatives and a decrease in percentage of heads in the older ages, presumably representing retirement and the loss of headship it usually entails. There does not appear to be a consistent temporal pattern in these trends over the six census listings.

For women, the age-related trends are a somewhat different (Figures 2-24, 2-26, 2-28, 2-

30, 2-32, 2-34). Instead of an increase in the percentage of heads with age, there is an increase in the percentage living as the spouse of the head. Only in later ages does the percentage of heads increase, a probable response to widowhood, which sometimes entails assuming headship. It may also be the case that widows also live as relatives (e.g. mothers, adult sisters), as the percentage living as relatives increases at higher ages, while the percentage living as spouses decreases. As with males, there is also a peak in the percentage living as non-kin that corresponds with life- cycle patterns in service.

11 See Chapter 4 for a detailed discussion of the practice of life-cycle service.

46 Similar patterns are seen in headship rates (Figure 2-35). Males assume nearly universal headship by their late 40‘s (approximately 90 percent of males of those ages are heads at each census enumeration). Females are much less likely to become heads of household. However, there is an increasing trend over time, probably related to the practice of widows assuming headship after the death of their husbands. The slight decrease in headship in old age observed in both sexes (although not for women in 1851 and 1861) is a possible sign of retirement, which usually requires giving up headship to an heir.

Unlike kin status, there appears to be little change in the percentage of individuals living in various household types over the life course, with the exception of an increase in the percentage of females (and to a lesser extent, males) living in extended households at the oldest ages (Figures 2-36, 2-37, 2-38). The percentage of both males and females living in simple, extended, and compound households is also broadly consistent across census years. Extended households, which include kin other than the head, his or her spouse, and their offspring, are the least common in northern Orkney, ranging from 17 to 22 percent of all sampled households

(Table 2-1). Compound and simple family households are more frequent, comprising 30-44 percent and 38-47 percent of households, respectively. Household types also vary in size.

Compound households are the largest, with an average household size of about 11 people.

Extended households include about six individuals on average, while simple households average about five members (Table 2-11). These household types also differ in mean and median valuation, a proxy measure of landholding size and quality. Compound households have the highest average valuation (Tables 2-12 and 2-13). Much of this higher average may be attributable to a few very large farms, which often contained several dwelling units, and were therefore, by definition, compound households, but likely functioned differently from smaller scale holdings farmed by related individuals. Indeed, the distribution of valuation is highly skewed (Figure 2-39).

47 Before households were categorized as simple, extended, or compound, they were classified using the Hammel-Laslett (1974) system, which has been adopted by many historical researchers, as it offers an exhaustive, although complicated, method of classification. For the purposes of this study, the Hammel-Laslett system was slightly modified to accommodate compound households. Each segment of a compound household, with its own head, was classified as if it were independent, but tagged to indicate that it belonged to a compound household. Tables 2-14 through 2-15 show the results of this classification. Solitary, no family, and simple family households are grouped as simple households for analysis in this study, and extended (at least one non-nuclear kin member present) and multiple family (more than one married couple) households are grouped as extended households. Compound households contain at least two segments, which of themselves can be either simple or extended in this classification system.

The proportion of households in each category can be compared to results from other historical samples to which this classification system has also been applied. If non-compound and the constituent segments of compound households are considered separately, the proportion of households classified into each family typology is comparable to proportions found in other regions of Northwest Europe, including England, Germany, and France (Table 2-20). If non- compound households are grouped hypothetically with the segments of compound households, as if compound households were not recognized, the proportions of households that fall into each family typology remain largely unchanged (Table 2-14 through 2-19).

However, if compound households are considered a kind of multiple-family household,

Orkney no longer conforms to the pattern of Northwest Europe. Rather, compound-family households inflate the numbers of multiple households (and by definition, complex households, which are the sum of extended and multiple households), so that Orkney appears more like

Southern and Eastern Europe (including Russia) than England or France (Table 2-21). Whatever

48 it is, something is going on in Orkney that does not conform to the classic interpretation of households in Northwest Europe promulgated by the Cambridge Group for the History of

Population and Social Structure (Laslett 2000; Laslett et al. 1983; Laslett and Wall 1974).

Discussion

This study highlights the usefulness of multiple data sources in the reconstruction of households in the past. The availability of information that complements census records and vital registers has led to the identification of the compound household, which would otherwise go unnoticed and result in the misinterpretation of census returns. For each case in which a compound household has been identified in the census, and sufficient supplementary information from historical maps and/or archaeological surveys exists, multiple dwellings with a single set of farm buildings have been indentified on the landscape of the corresponding farmstead. This suggests that the inhabitants of compound households were likely to have cooperated in the operation of the farm. Further, in the majority of cases, kinship ties among the component units of compound households can be established. However, in a few examples, no close kinship ties can be found. These instances, in which apparently unrelated families share a farmstead, demonstrate diversity in household forms, but should be interpreted cautiously, as they do not conform to commonly accepted kinship-based definitions of households. In addition, it is important to remember that kinship relationships are relatively easy to reconstruct given the types of information available, while economic behavior, such as cooperative farming and resource pooling, can only be inferred.

With these caveats in mind, if compound households can be taken to be roughly equivalent to the definition of complex households employed by historical demographers (the combined frequency of multiple-family and extended-family households), then northern Orkney

49 was characterized by higher levels of household complexity than would be expected for northwest Europe. This pattern may be attributable to one or more characteristics of the islands.

First, inheritance in Orkney is predominantly impartible, meaning that the farmstead is passed on to the next generation largely intact, rather than divided up among several heirs. Impartible inheritance is often observed in conjunction with stem-family systems, in which a single heir and his family coreside with parents and inherit upon their retirement or death. In several of the case studies described above, stem-family arrangements are evident. This type of property transfer could contribute to higher levels of household complexity (in this case, three-generation households) than would be found in household formation systems that require neolocal postmarital residence (For descriptions of such systems, see Engelen and Wolf 2005; Hajnal

1965; Hajnal 1982).

Another possible interpretation of the high instance of complex households in Orkney comes from some ideas put forth in a study of Hungarian villages (Andorka and Balazs-Kovacs

1986). The authors hypothesize that growing household complexity was a response to the growing scarcity of land. In fact, complex households are often found in contexts where individual households have little control over the size of their landholdings, and the establishment of new holdings and farmsteads is strongly discouraged, such as with sharecroppers in Italy or serfs in several regions of Eastern Europe (Czap 1982; Czap 1983; Kertzer 1984; Szoltysek

2007). In Orkney, land was primarily owned and disposed of by landholders known as lairds.

Lairds held large tracts of land and were usually absent from the Northern Isles, while a factor, steward, or trust administered property in their stead. These landholders made important decisions about the size of holdings, the amount of rent, and the implementation of improvements such as drainage (Schrank and Lockhart 1995).

While social and economic relationships influence the availability of land, so do ecological constraints. In fact, several studies have stressed the need to recognize environmental

50 factors in conjunction with socioeconomic conditions (Kahk et al. 1982; Mitterauer 1992; Moring

1999). Ecological conditions in Orkney may also contribute to the high frequency of complex households. Land is an important limiting resource in this well-populated island environment, so households may become more complex in response to ecologically-based land shortages, rather than limitations imposed by landlords (in the case of sharecroppers) or seigneurs (in the case of serfs). However, the effects of landlords and ecology may, in practice, be difficult to separate. In fact, it has been noted that the islands of Estonia had higher frequencies of multiple-family households than the northern or southern regions of that country, even though serfdom and corvée labor requirements, which tend to increase household size and complexity, dominated the mainland economy (Kahk et al. 1982). However, the exact nature of this potential island effect and its contribution relative to that of impartible inheritance and other socioeconomic factors on the frequency of complex households in Orkney remains a topic for future investigation.

Finally, this descriptive study demonstrates that households are dynamic entities whose composition and size change over time. Households should therefore not be taken as static groups or contexts in which individuals function. Rather, the causes and consequences of household change should be considered in their own rights. Chapter 3 describes one way that the study of household change may be accomplished.

Further, the hypothetical comparison of the proportions of household types if compound households were ignored has some interesting consequences. If compound households are ignored, the proportions of simple, complex, and multiple households in the population does not appreciably change. This observation suggests that from census data and proportions of household types alone, it is not easy to tell whether compound households existed. Thus, it is possible that some accounts of household forms in Northwest Europe that use only census listings may underestimate household complexity, not only through cross-sectional nature of the data, but also in the interpretation of household groupings. In other words, compound households may

51 have existed elsewhere in Northwest Europe, but they have been ―hidden‖ because historical records in isolation may not be able to uncover them.

52 Figures and Tables

Table 2-1. Percentage of households by type and census year. Percentages are rounded. Compound Extended Simple N 1851 42.5 17.4 40.6 598 1861 44.0 17.4 38.4 625 1871 38.5 20.5 41.1 649 1881 39.9 21.2 38.9 684 1891 35.1 19.8 45.1 650 1901 30.8 22.1 47.1 647

Table 2-2. The inhabitants of South Hammer by census year. Data presented in this (and similar tables) present information as listed in census returns, including obvious inconsistencies (in ages, for example). Genealogy Relationship to Year of Number Name Head Age Birth Occupation Inhabitants of South Hammer, 1851 South Hammer 6 Ann Paterson Daughter 6 1845 South Hammer 2 Ann Paterson Wife 40 1811 South Hammer 1 Charles Paterson Head 41 1810 Farmer (10 acres employs 2 laborers) South Hammer 7 George Paterson Son 4 1847 South Hammer 9 Janet Paterson Daughter 1 1850 South Hammer 5 Mary Paterson Daughter 10 1841 South Hammer 4 Stewart Paterson Son 14 1837 South Hammer 8 William Paterson Son 2 1849

Inhabitants of South Hammer, 1861 South Hammer 6 Ann Paterson Daughter 16 1845 South Hammer 2 Ann Paterson Wife 50 1811 South Hammer 1 Charles Paterson Head 51 1810 Farmer South Hammer 7 George Paterson Son 14 1847 South Hammer 11 James Paterson Son 8 1853 South Hammer 12 Jean Paterson Daughter 4 1857 South Hammer 10 Jessie Paterson Daughter 10 1851 Scholar South Hammer 5 Mary Paterson Daughter 20 1841 South Hammer 4 Stewart Paterson Son 24 1837 Blacksmith South Hammer 8 William Paterson Son 12 1849 Scholar

Inhabitants of South Hammer, 1871 South Hammer 1 15 Mary Logie Step-Daughter 11 1860 Scholar South Hammer 1 16 Charles Paterson Son 6 1865 Scholar South Hammer 1 17 Jessieann Paterson Daughter 3 1868 South Hammer 1 18 John David Paterson Son 2 1869 South Hammer 1 12 Lydia Paterson Wife 38 1833 South Hammer 1 4 Stewart Paterson Head 36 1835 Blacksmith

South Hammer 2 2 Ann Paterson Wife 60 1811 South Hammer 2 1 Charles Paterson Head 61 1810 Farmer South Hammer 2 9 Jannet Paterson Daughter 21 1850 South Hammer 2 12 Jean Paterson Daughter 14 1857

South Hammer 3 15 Isabella Paterson Wife 27 1844 South Hammer 3 8 William Paterson Head 22 1849 Blacksmith

53

Table 2-2. Continued.

Inhabitants of South Hammer, 1881 South Hammer 1 2 Ann Paterson Wife 70 1811 South Hammer 1 1 Charles Paterson Head 71 1810 Farmer South Hammer 1 12 Jane Paterson Daughter 23 1858 General domestic servant

South Hammer 2 19 Ann Paterson Daughter 9 1872 Scholar South Hammer 2 16 Charles Paterson Son 16 1865 Blacksmith Apprentice South Hammer 2 18 John D. Paterson Son 12 1869 Scholar South Hammer 2 14 Lydia Logie Paterson Wife 48 1833 South Hammer 2 20 Robert Paterson Son 6 1875 Scholar South Hammer 2 4 Stewart Paterson Head 44 1837 Master Blacksmith South Hammer 2 21 Stewart Paterson Son 4 1877

South Hammer 3 15 Isabella Paterson Wife 37 1844 South Hammer 3 24 Robert Paterson Son 9 1872 South Hammer 3 8 William Paterson Head 32 1849 Fisherman

Inhabitants of South Hammer, 1891 South Hammer 1 15 Isabella Paterson Wife 47 1844 South Hammer 1 24 Robert Paterson Son 19 1872 Fisherman South Hammer 1 8 William Paterson Head 42 1849 Fisherman South Hammer 1 N/A James R. Pottinger Boarder 7 1884 Scholar

South Hammer 2 14 Lydia Paterson Wife 57 1834 South Hammer 2 4 Stewart Paterson Head 53 1838 Blacksmith South Hammer 2 21 Stewart Paterson Son 14 1877 Blacksmith Apprentice

South Hammer 3 22 George Paterson Son 22 1869 Farm Servant South Hammer 3 13 Mary Paterson Wife 48 1843 South Hammer 3 3 Robert Paterson Head 56 1835 Farmer South Hammer 3 23 Robert Paterson Son 15 1876 Farm Servant

Inhabitants of South Hammer, 1901 South Hammer 1 27 Charles Rendall Son 13 1888 Scholar South Hammer 1 9 Janet Rendall Head 51 1850 Housekeeper South Hammer 1 28 Jessie Rendall Daughter 11 1890 Scholar South Hammer 1 26 John Rendall Son 17 1884 Ploughman on Farm South Hammer 1 25 William Rendall Son 21 1880 Ploughman on Farm

South Hammer 2 13 Mary Paterson Wife 58 1843 South Hammer 2 3 Robert Paterson Head 66 1835 Farmer South Hammer 2 23 Robert Paterson Son 24 1877 Assisting on Farm

South Hammer 3 15 Isabella Paterson Wife 56 1845 South Hammer 3 24 Robert Paterson Son 29 1872 Fisherman South Hammer 3 8 William Paterson Head 52 1849 Fisherman

54 Figure 2-1. Genealogical Relationships among the inhabitants of South Hammer.

55 Figure 2-2. Historical and modern maps of South Hammer with building types labeled.

Dwelling Dwelling Dwelling Dwelling Dwelling Dwelling

Barn & Kiln Barn & Kiln

Stable Stable Smithy Smithy Pig House Byre Pig House Byre

56 Figure 2-3. Modern photograph of the South Hammer croft complex. Photograph courtesy of Dr. James Wood

57 Table 2-3. Inhabitants of Sangar by census year.

Genealogy Relationship Year of Number Name to Head Age Birth Occupation Inhabitants of Sangar, 1851 Sagar 3 Barbara Drever Daughter 11 1840 Scholar Sagar 2 Bili Drever Wife 36 1815 Sagar 4 David Drever Son 7 1844 Scholar Sagar 5 Margaret Drever Daughter 5 1846 Scholar Sagar 1 Thomas Drever Head 43 1808 Farmer (15 acres) Sagar 6 William Drever Son 3 1848

Inhabitants of Sangar, 1861 Sanger 3 Barbara Drever Daughter 21 1840 Sanger 2 Isabella Drummond Drever Wife 45 1816 Sanger 7 Jane Drever Daughter 8 1853 Sanger 8 Mary Drever Daughter 2 1859 Sanger 1 Thomas Drever Head 53 1808 Farmer Sanger 6 William Drever Son 13 1848

Inhabitants of Sangar, 1871 Sanger 2 Isabella Drever Wife 56 1815 Sanger 7 Jean Drever Daughter 18 1853 Sanger 8 Mary Drever Daughter 12 1859 Sanger 1 Thomas Drever Head 63 1808 Farmer Sanger 6 William Drever Son 23 1848 Fisherman

Inhabitants of Sangar, 1881 Sanger No1 2 Bell Drever Wife 66 1815 Sanger No1 7 Jean Drever Daughter 28 1853 Sanger No1 8 Mary Drever Daughter 22 1859 Sanger No1 1 Thomas Drever Head 73 1808 Farmer

Sanger No2 9 Betsy Drever Wife 29 1852 Sanger No2 10 David Drever Son 6 1875 Scholar Sanger No2 12 Elizabeth Drever Daughter 2 1879 Sanger No2 6 William Drever Head 33 1848 Sailor, merchant service Sanger No2 11 William Drever Son 4 1877

58

Table 2-3. Continued

Inhabitants of Sangar, 1891 Sanger No1 9 Betsy Drever Wife 38 1853 Sanger No1 12 Elizabeth Drever Daughter 12 1879 Sanger No1 14 Isabella Drever Daughter 2 1889 Sanger No1 14 John Drever Son 3 1888 Sanger No1 13 Peter Drever Son 8 1883 (Dumb) Sanger No1 6 William Drever Head 43 1848 Farmer

Sanger No2 2 Isabella Drever Wife 76 1815 Sanger No2 8 Mary Drever Daughter 32 1859 Sanger No2 1 Thomas Drever Head 83 1808 Farmer

Inhabitants of Sangar, 1901 Sangar 14 Bella Drever Daughter 12 1889 Scholar Sangar 9 Betsy Drever Wife 49 1852 Sangar 16 James Drever Son 9 1892 Scholar Sangar 17 Thomas Drever Son 7 1894 Scholar Sangar 6 William Drever Head 53 1848 Farmer, Horseman

Figure 2-4. Genealogical relationships among the inhabitants of Sangar.

59 Figure 2-5. Historical and modern maps of Sangar.

Barn & Kiln Dwelling

Byre Dwelling

Barn & Kiln Dwelling Dwelling converted to Byre/Stable Byre Dwelling Hen & Pig House

60 Figure 2-6. Modern photograph of the Sangar croft complex. Photograph by the author.

61 Table 2-4. Inhabitants of Quoynabreckan by census year.

Genealogy Relationship Year of Number Name to Head Age Birth Occupation Inhabitants of Quoynabreckan, 1851 Queen of Brecken 8 David Rendall Son 1 1850 Queen of Brecken 3 James Rendall Son 16 1835 Scholar Queen of Brecken 1 John Rendall Head 41 1810 Farmer (5 acres employs 1 laborer) Queen of Brecken 7 John Rendall Son 6 1845 Queen of Brecken 5 Margaret Rendall Daughter 10 1841 Scholar Queen of Brecken 2 Margaret Rendall Wife 43 1808 Queen of Brecken 4 William Rendall Son 16 1835 Farm servant

Inhabitants of Quoynabreckan, 1861 Quinabreckin 8 David Rendall Son 11 1850 Quinabreckin 11 John Rendall Grandson 3 1858 Quinabreckin 1 John Rendall Head 50 1811 Farmer Quinabreckin 7 John Rendall Son 16 1845 Quinabreckin 6 Louisa Rendall Daughter 19 1842 Quinabreckin 2 Margaret Lennie Rendall Wife 52 1809

Inhabitants of Quoynabreckan, 1871 Queenabreckin 10 Isabella Rendall Wife 27 1844 Queenabreckin 7 John Rendall Head 26 1845 Farmer Queenabreckin 2 Margaret Rendall Mother 64 1807 Shareholder of farm Queenabreckin 12 William Rendall Son 0 1871 Queenabreckin N/A Louisa Swanney Servant 28 1843 Farm servant domestic

Inhabitants of Quoynabreckan, 1881 Queenabreckan 15 Ann Rendall Daughter 4 1877 Queenabreckan 13 Charles Rendall Son 8 1873 Scholar Queenabreckan 10 Isabella Rendall Wife 36 1845 Queenabreckan 16 James Rendall Son 3 1878 Queenabreckan 7 John Rendall Head 35 1846 Farmer Queenabreckan 14 John Rendall Son 7 1874 Scholar Queenabreckan 17 Margaret Rendall Daughter 1 1880 Queenabreckan 12 William Rendall Son 10 1871 Scholar

Inhabitants of Quoynabreckan, 1891 Queenobreckan 15 Ann Rendall Daughter 15 1876 General Domestic Servant Queenobreckan 13 Charles Rendall Son 18 1873 Queenobreckan 20 George Rendall Son 1 1890 Queenobreckan 10 Isabella Rendall Wife 47 1844 Queenobreckan 16 James Rendall Son 13 1878 Scholar Queenobreckan 19 Jane Rendall Daughter 5 1886 Queenobreckan 7 John Rendall Head 46 1845 Farmer Queenobreckan 18 Louisa Rendall Daughter 7 1884 Scholar Queenobreckan 17 Margaret Rendall Daughter 12 1879 Scholar Queenobreckan 12 William Rendall Son 20 1871 Fisherman

Inhabitants of Quoynabreckan, 1901 Quoynabrackan 20 George Rendall Son 11 1890 Scholar Quoynabrackan 10 Isabella Rendall Wife 57 1844 Quoynabrackan 16 James Rendall Son 22 1879 (W/H) Quoynabrackan 19 Jane Rendall Daughter 15 1886 General Servant Domestic Quoynabrackan 7 John Rendal Head 56 1845 Farmer

62 Figure 2-7. Genealogical relationships among the inhabitants of Quoynabreckan.

63 Figure 2-8. Historical and modern maps of Quoynabreckan.

64 Figure 2-9. Modern Photograph of the Quoynabreckan croft complex. Photograph courtesy of Dr. James Wood.

65 Table 2-5. Inhabitants of Ha‘Ouse by census year.

Genealogy Relationship Year of Number Name to Head Age Birth Occupation Inhabitants of Ha'Ouse, 1851 Ha'ouse 7 Barbra Rousay Daughter 4 1847 Ha'ouse 3 John Rousay Jr. Head 27 1824 Fisherman Ha'ouse 6 Louisa (Elizabeth) Wilson Rousay Wife 29 1822 Ha'ouse 8 Peter Rousay Son 1 1850

Ha'ouse2 2 Helen Manson Rousay Wife 57 1794 Ha'ouse2 5 Ann Rousay Daughter 15 1836 Ha'ouse2 1 John Rousay Sr. Head 59 1792 Farmer (4 acres), employs 1 girl

Inhabitants of Ha'Ouse, 1861 Ha Ouse 7 Barbara Rousay Daughter 14 1847 Scholar Ha Ouse 6 Elizabeth Rousay Wife 40 1821 Ha Ouse 9 John Rousay (III) Son 7 1854 Scholar Ha Ouse 3 John Rousay (Jr.) Head 39 1822 Farmer Ha Ouse 8 Peter Rousay Son 11 1850 Scholar

Ha Ouse 2 1 John Rousay (Sr.) Head 70 1791 Farmer Ha Ouse 2 4 Maragaret Rousay Daughter 28 1833 Housekeeper

Inhabitants of Ha'Ouse, 1871 Hallouse 1 John Rousay (Sr.) Head 80 1791 Farmer Hallouse 4 Margaret Rousay Daughter 37 1834

Hallouse 7 Barbara Rousay Daughter 24 1847 Hallouse 6 Elisabeth Rousay Wife 48 1823 Hallouse 3 John Rousay (Jr.) Head 47 1824 Farmer Hallouse 9 John Rousay (III) Son 17 1854

Inhabitants of Ha'Ouse, 1881 Hawhouse 7 Barbara Rousay Daughter 34 1847 Hawhouse 6 Elizabeth Rousay Wife 57 1824 Hawhouse 3 John Rousay Head 57 1824 Farmer Hawhouse 9 John Rousay Son 26 1855

Hawhouse 4 Margaret Rousay Head 48 1833 Spinner

Inhabitants of Ha'Ouse, 1891 Halhouse 7 Barbara Rousay Daughter 44 1847 Dressmaker Halhouse 3 John Rousay Head 67 1824 Farmer Halhouse 9 John Rousay Son 36 1855 Halhouse N/A Ann Scott Servant 16 1875 General Domestic Servant

Halhouse No2 4 Margaret Rousay Lodger 57 1834 Woolen spinner

Inhabitants of Ha'Ouse, 1901 Hahouse 1 N/A Margaret Lennie Servant 53 1848 General Domestic Hahouse 1 7 Barbra Rousay Daughter 54 1847 Housekeeper Hahouse 1 3 John Rousay Head 76 1825 Farmer Hahouse 1 9 John Rousay Son 47 1854 Assisting on Farm

Hahouse 2 4 Margaret Rousay Head 66 1835 Wool Spinner

66

Figure 2-10. Genealogical relationships among the inhabitants of Ha‘Ouse.

67 Figure 2-11. Historic and modern maps of Ha‘Ouse.

68 Figure 2-12. Modern photograph of the Ha‘Ouse croft complex. Photography courtesy of Dr. James Wood.

69 Table 2-6. Inhabitants of Stanegeo by census year.

Genealogy Relationship Year of Number Name to Head Age Birth Occupation Inhabitants of Stanegeo, 1881 Stangeo 2 Betsy (Pottinger) Rendall Wife 35 1846 Stangeo 1 David Rendall Head 31 1850 Farmer Stangeo 7 David Rendall Son 1 1880 Stangeo 5 James Rendall Son 6 1875 Scholar Stangeo 3 Joann Rendall Daughter 10 1871 Scholar Stangeo 6 Margaret Rendall Daughter 3 1878 Stangeo 4 Thomas Rendall Son 9 1872 Scholar

Inhabitants of Stanegeo, 1891 Stanegoe 2 Betsy Rendall Wife 45 1846 Stanegoe 1 David Rendall Head 41 1850 Farmer Stanegoe 7 David Rendall Son 10 1881 Scholar Stanegoe 11 Elizabeth Rendall Daughter 0 1891 Stanegoe 9 George Rendall Son 5 1886 Stanegoe 5 James Rendall Son 16 1875 Stanegoe 6 Margaret D. Rendall Daughter 13 1878 Scholar Stanegoe 10 Sinclair Rendall Son 2 1889 Stanegoe 4 Thomas Rendall Son 19 1872 Fisherman Stanegoe 8 William Rendall Son 8 1883 Scholar

Inhabitants of Stanegeo, 1901 Stanegoe 1 2 Betsy Rendall Wife 55 1846 Stanegoe 1 12 Catherine Rendall Daughter 7 1894 Scholar Stanegoe 1 1 David Rendall Head 51 1850 Farmer & Kelpmaker Stanegoe 1 11 Elizabeth Rendall Daughter 10 1891 Scholar Stanegoe 1 9 George Rendall Son 15 1886 Fisherman & Kelpmaker Stanegoe 1 6 Margaret Rendall Daughter 23 1878 Domestic Work Stanegoe 1 16 Robert Rendall Grandson 5 1896 Scholar Stanegoe 1 10 Sinclair Rendall Son 12 1889 Scholar Stanegoe 1 4 Thomas Rendall Son 29 1872 Fisherman & Kelpmaker

Stanegoe 2 14 David Rendall Son 2 1899 Stanegoe 2 15 Elizabeth Rendall Daughter 0 1901 Stanegoe 2 5 James Rendall Head 26 1875 Fisherman & Kelpmaker Stanegoe 2 13 Jessie (Drever) Rendall Wife 30 1871

70 Figure 2-13. Genealogical relationships among the inhabitants of Stanegeo.

71 Figure 2-14. Historical and modern maps of Stanegeo.

72 Figure 2-15. Modern Photograph of the Stanegeo croft complex. Photograph courtesy of Dr. James Wood.

73 Table 2-7. The inhabitants of Nether Brough by census year.

Genealogy Relationship to Year of Number Name Head Age Birth Occupation Inhabitants of Nether Brough, 1851 Nether Brugh Alexander Drever Son 2 1849 Nether Brugh Ann Drever Daughter 0 1851 Nether Brugh George Drever Son 16 1835 Farm Servant Nether Brugh Helen Drever Wife 45 1806 Nether Brugh Isabella Drever Daughter 22 1829 Farm Servant Nether Brugh Jane Drever Daughter 5 1846 Scholar Nether Brugh John Drever Head 46 1805 Fisherman Nether Brugh William Drever Son 11 1840 Scholar

Nether Brugh2 2 Elizabeth Rendall Wife 46 1805 Nether Brugh2 11 George Rendall Son 3 1848 Nether Brugh2 6 Isabella Rendall Daughter 15 1836 Scholar Nether Brugh2 5 Jean Rendall Daughter 17 1834 Nether Brugh2 9 Jeremiah Rendall Son 9 1842 Scholar Nether Brugh2 7 Louisa Rendall Daughter 12 1839 Scholar Nether Brugh2 1 Stewart Rendall Head 49 1802 Hand loom weaver and farmer, employs 1 girl Nether Brugh2 10 William Rendall Son 6 1845 Scholar

Inhabitants of Nether Brough, 1861 Neather Brough Alexander Drever Son 12 1849 Scholar Neather Brough Ann Drever Daughter 10 1851 Scholar Neather Brough Hellen Drever Wife 54 1807 Neather Brough James Drever Son 29 1832 Tailor Neather Brough Jane Drever Daughter 16 1845 Neather Brough John Drever Head 55 1806 Fisherman Neather Brough Jane Foulis Mother 79 1782

Neather Brough 2 2 Elizabeth Rendall Wife 54 1807 Neather Brough 2 11 George Rendall Son 13 1848 Scholar Neather Brough 2 9 Jeremiah Rendall Son 20 1841 Neather Brough 2 1 Stewart Rendall Head 58 1803 Fisherman Neather Brough 2 10 William Rendall Son 16 1845

Inhabitants of Nether Brough, 1871 Nether Brough 2 Elizabeth Rendall Wife 65 1806 Nether Brough 11 George Rendall Son 22 1849 Woolen weaver Nether Brough 9 Jeremiah Rendall Son 29 1842 Woolen weaver Nether Brough 1 Stewart Rendall Head 68 1803 Woolen weaver Nether Brough 10 William Rendall Son 26 1845 Blacksmith Nether Brough 15 James Smith Grandson 9 1862 Scholar

Nether Brough 14 Elizabeth McGregor Niece 16 1855 General servant domestic Nether Brough 16 Elizabeth Smith Daughter 5 1866 Nether Brough 5 Jane Smith Head 36 1835 Sewing Teacher

74 Table 2-7. Continued.

Inhabitants of Nether Brough, 1881 Netherbrough 14 Bessy Macgregor Granddaughter 27 1854 General domestic servant Netherbrough 2 Elizabeth Rendall Wife 75 1806 Netherbrough 9 Jeremiah Rendall Son 40 1841 Fisherman Netherbrough 1 Stewart Rendall Head 78 1803 Woolen weaver master

Netherbrough 11 George Rendall Head 33 1848 Woolen weaver master Netherbrough 17 George Smith Rendall Son 4 1877 Netherbrough 18 Henry McRobert Rendall Son 1 1880 Netherbrough 13 Margaret Rendall Wife 28 1853

Netherbrough 16 Elizabeth M. Smith Daughter 15 1866 Assistant dressmaker Netherbrough 5 Jane Smith Head 46 1835 Dressmaker

Inhabitants of Nether Brough, 1891 Nether Brough 14 Elizabeth Rendall Niece 37 1854 General Domestic Servant Nether Brough 19 Elizabeth S. Rendall Daughter 8 1883 Scholar Nether Brough 21 Fanny S. Rendall Daughter 1 1890 Nether Brough 11 George Rendall Head 43 1848 Woolen weaver Nether Brough 17 George S. Rendall Son 14 1877 Scholar Nether Brough 18 Henry McD. Rendall Son 11 1880 Scholar Nether Brough 9 Jeremiah Rendall Brother 50 1841 Woolen weaver Nether Brough 13 Margaret Rendall Wife 38 1853 Nether Brough 20 Stewart Rendall Son 5 1886 Scholar

Inhabitants of Nether Brough, 1901 Netherbrough 14 Betsy McGregor Niece 47 1854 General Servant Domestic Netherbrough 21 Fanny Rendall Daughter 11 1890 Scholar Netherbrough 11 George Rendall Head 53 1848 Woollen Weaver Netherbrough 18 Henry McRobert Rendall Son 21 1880 Journeyman Tailor Netherbrough 23 Jane Rendall Daughter 5 1896 Scholar Netherbrough 9 Jeremiah S. Rendall Brother 60 1841 Woollen Weaver Netherbrough 13 Margaret Rendall Wife 48 1853 Netherbrough 22 Robert Rendall Son 7 1894 Scholar

75 Figure 2-16. Genealogical relationships among the inhabitants of Nether Brough.

76 Figure 2-17. Historical and modern maps of Nether Brough.

77 Figure 2-18. Modern photograph of the Nether Brough croft complex, southern block. Photograph courtesy of Dr. Tim Murtha.

Figure 2-19. Modern photograph of the Nether Brough croft complex, northern block. Photograph courtesy of Dr. James Wood.

78 Table 2-8. Inhabitants of Moa by census year.

Genealogy Relationship to Year of Number Name Head Age Birth Occupation Inhabitants of Moa, 1851 Moah 10 Ann Drever Wife 34 1817 Moah 22 Barbara Drever Daughter 0 1851 Moah 20 Francis Drever Daughter 6 1845 Scholar Moah 21 John Drever Son 2 1849 Moah 19 Robert Drever Son 7 1844 Scholar Moah 6 William Drever Head 34 1817 Farmer (20 Acres)

Moah2 12 Barbara Dearness Granddaughter 21 1830 Seamstress Moah2 13 James Dearness Grandson 19 1832 Moah2 2 Mary Drever Daughter 49 1802 Housekeeper Moah2 1 Robert Drever (Sr.) Head 77 1774 Farmer (20 Acres, employs 1 laborer)

Moah3 8 Margaret Drever Wife 26 1825 Moah3 14 Robert Drever Son 4 1847 Moah3 5 Robert Drever (Jr.) Head 38 1813 Farmer (20 Acres)

Inhabitants of Moa, 1861 Moah 1 22 Barbara Drever Daughter 10 1851 Moah 1 20 Francis Drever Daughter 15 1846 Moah 1 11 Isabelle Drever Wife 37 1824 Moah 1 21 John Drever Son 12 1849 Moah 1 19 Robert Drever Son 18 1843 Moah 1 6 William Drever Head 44 1817 Farmer and boatbuilder

Moah 2 9 Ann Drever Wife 36 1825 Moah 2 15 John Drever Son 1 1860 Moah 2 5 Robert Drever Head 50 1811 Farmer Moah 2 14 Robert Drever Son 13 1848

Moah 3 N/A Isabella Seater Head 41 1820 Formerly servant

Inhabitants of Moa, 1871 Moa 23 Ann Drever Daughter 9 1862 Moa 22 Barbra Drever Daughter 20 1851 Moa 24 Isabella Drever Daughter 4 1867 Moa 11 Isabella Drever Wife 47 1824 Moa 25 Jean Drever Daughter-in-law 22 1849 Moa 21 John Drever Son 22 1849 Joiner Moa 19 Robert Drever Son 28 1843 Blacksmith Moa 6 William Drever Head 54 1817 Farmer

79 Table 2-8. Continued.

Inhabitants of Moa, 1881 Moa 23 Ann Drever Daughter 19 1862 Moa 27 Ann Drever Granddaughter 7 1874 Scholar Moa 11 Bell Drever Wife 57 1824 Moa 26 Elizabeth Drever Granddaughter 9 1872 Scholar Moa 24 Isabell Drever Daughter 14 1867 Scholar Moa 25 Jean Drever Daughter-in-law 37 1844 Moa 19 Robert Drever Son 39 1842 Joiner (master) Moa 28 William Drever Grandson 3 1878 Moa 6 William Drever Head 64 1817 Farmer

Moa 9 Ann Drever Wife 59 1822 Moa 16 George Drever Son 19 1862 Fisherman Moa 17 James Drever Son 17 1864 Laborer Moa 15 John Drever Son 21 1860 Laborer Moa 18 Margaret Drever Daughter 9 1872 Scholar Moa 5 Robert Drever Head 69 1812 Farm Laborer

Inhabitants of Moa, 1891 Moah No1 27 Ann Drever Granddaughter 17 1874 General Domestic Servant Moah No1 23 Ann Drever Daughter 28 1863 General Domestic Servant Moah No1 29 MaryAnn Drever Granddaughter 9 1882 Scholar Moah No1 19 Robert Drever Son 48 1843 Boat Builder Moah No1 28 William Drever Grandson 13 1878 Scholar Moah No1 6 William Drever Head 74 1817 Farmer

Moah No2 9 Ann Drever Wife 60 1831 Moah No2 16 George Drever Son 29 1862 Moah No2 17 James Drever Son 27 1864 Moah No2 15 John Drever Son 30 1861 Moah No2 18 Maggie Drever Daughter 22 1869 Moah No2 5 Robert Drever Head 79 1812 Farmer

Inhabitants of Moa, 1901 Moah No.1 29 Mary A. Drever Granddaughter 19 1882 Domestic Servant Moah No.1 19 Robert Drever Son 58 1843 Joiner & Carpenter Moah No.1 28 William Drever Grandson 23 1878 Joiner & horseman on farm Moah No.1 6 William Drever Head 84 1817 Farmer

Moah No. 2 9 Ann Drever Head 74 1827 Moah No. 2 16 George Drever Son 39 1862 Boatman Moah No. 2 15 John Drever Son 40 1861 Horseman Moah No. 2 18 Margaret Drever Daughter 32 1869 Domestic Servant

80

Figure 2-20. Genealogical relationships among the inhabitants of Moa.

24

23

1882

-

11

1827

22

6

21

20

1855

-

10

1823

29

188?

-

25

1849

9

28

19

1896

-

27

5

1810

18

26

1857

-

8

17

1824

4

16

15

3

1859

-

1

14

7

1774

13

2 12

81 Figure 2-21. Historical and modern maps of Moa.

82 Figure 2-22. Modern photograph of the Moa croft complex. Photography courtesy of Dr. James Wood.

83 Table 2-9. Percentage of sample by household kin status relative to head or head‘s spouse. For example, ―Mother‖ includes both the head‘s mother and mother-in-law. 1851 1861 1871 1881 1891 1901 Nuclear* 80.8 80.5 81.2 81.8 82.3 79.7 Father 0.3 0.4 0.4 0.5 0.2 0.1 Mother 1.4 1.3 1.3 1.4 1.4 2.0 Sister 1.7 1.6 1.9 2.2 1.7 2.6 Brother 0.6 0.7 0.7 0.5 0.8 1.2 Nephew 0.5 0.4 0.4 0.6 0.4 0.8 Niece 0.6 0.3 0.5 0.5 0.6 0.7 Grandchild 2.9 2.9 3.8 4.3 4.3 4.9 Son-in-law 0.5 0.3 0.4 0.5 0.5 0.5 Daughter-in-law 0.4 0.5 0.7 0.7 0.7 0.7 Other Kin 0.6 2.1 1.1 1.3 0.9 1.7 Non-Kin 9.7 8.9 7.4 5.7 6.1 5.2

*Head, Spouse, Children

Table 2-10. Percentage of sampled households that include kin, reckoned relative to the head or head‘s spouse. For example, ―Mother‖ includes both the head‘s mother and mother-in-law. 1851 1861 1871 1881 1891 1901 Nuclear* 99.5 99.8 99.4 100.0 98.8 98.8 Father 2.3 3.5 2.9 3.9 1.5 0.9 Mother 11.4 10.9 10.5 10.4 9.8 13.1 Sister 11.2 10.2 12.3 14.0 11.2 14.4 Brother 4.8 4.5 5.1 3.2 4.8 6.2 Nephew 3.0 2.9 3.2 3.9 3.1 4.5 Niece 4.2 2.4 3.7 3.4 4.0 3.7 Grandchild 13.4 14.6 16.3 17.0 16.6 17.2 Son-in-law 4.3 2.2 3.5 3.5 3.5 3.4 Daughter-in-law 3.5 4.0 5.2 5.3 4.8 4.6 Other Kin 4.2 5.1 7.4 7.2 5.8 8.2 Non-Kin 32.9 30.6 27.9 23.8 26.3 21.3

*Head, Spouse, Children

84 Figure 2-23. Percentage of sample by age and relationship to the household head.

Males, 1851 1 0.9 0.8 0.7 0.6 Children 0.5

Percent 0.4 Spouse 0.3 Head 0.2 0.1 Relative 0

Non-Kin

5

-

15 65 10 20 25 30 35 40 45 50 55 60 70

0

------

70+

5

10 60 15 20 25 30 35 40 45 50 55 65 Age Group

Figure 2-24. Percentage of sample by age and relationship to the household head.

Females, 1851 1 0.9 0.8 0.7 0.6 Children 0.5

Percent 0.4 Spouse 0.3 Head 0.2 0.1 Relative 0

Non-Kin

5

-

10

15 65 20 25 30 35 40 45 50 55 60 70

0 -

------

70+

5

10 60 15 20 25 30 35 40 45 50 55 65 Age Group

85 Figure 2-25. Percentage of sample by age and relationship to the household head.

Males, 1861 1 0.9 0.8 0.7 0.6 Children 0.5

Percent 0.4 Spouse 0.3 Head 0.2 0.1 Relative 0

Non-Kin

5

-

15 65 10 20 25 30 35 40 45 50 55 60 70

0

------

70+

5

10 60 15 20 25 30 35 40 45 50 55 65 Age Group

Figure 2-26. Percentage of sample by age and relationship to the household head.

Females, 1861 1 0.9 0.8 0.7 0.6 Children 0.5

Percent 0.4 Spouse 0.3 Head 0.2 0.1 Relative 0

Non-Kin

5

-

10

15 65 20 25 30 35 40 45 50 55 60 70

0 -

------

70+

5

10 60 15 20 25 30 35 40 45 50 55 65 Age Group

86 Figure 2-27. Percentage of sample by age and relationship to the household head.

Males, 1871 1 0.9 0.8 0.7 0.6 Children 0.5

Percent 0.4 Spouse 0.3 Head 0.2 0.1 Relative 0

Non-Kin

5

-

15 65 10 20 25 30 35 40 45 50 55 60 70

0

------

70+

5

10 60 15 20 25 30 35 40 45 50 55 65 Age Group

Figure 2-28. Percentage of sample by age and relationship to the household head.

Females, 1871 1 0.9 0.8 0.7 0.6 Children 0.5

Percent 0.4 Spouse 0.3 Head 0.2 0.1 Relative 0

Non-Kin

5

-

10

15 65 20 25 30 35 40 45 50 55 60 70

0 -

------

70+

5

10 60 15 20 25 30 35 40 45 50 55 65 Age Group

87 Figure 2-29. Percentage of sample by age and relationship to the household head.

Males, 1881 1 0.9 0.8 0.7 0.6 Children 0.5

Percent 0.4 Spouse 0.3 Head 0.2 0.1 Relative 0

Non-Kin

5

-

15 65 10 20 25 30 35 40 45 50 55 60 70

0

------

70+

5

10 60 15 20 25 30 35 40 45 50 55 65 Age Group

Figure 2-30. Percentage of sample by age and relationship to the household head.

Females, 1881 1 0.9 0.8 0.7 0.6 Children 0.5

Percent 0.4 Spouse 0.3 Head 0.2 0.1 Relative 0

Non-Kin

5

-

10

15 65 20 25 30 35 40 45 50 55 60 70

0 -

------

70+

5

10 60 15 20 25 30 35 40 45 50 55 65 Age Group

88 Figure 2-31. Percentage of sample by age and relationship to the household head.

Males, 1891 1 0.9 0.8 0.7 0.6 Children 0.5

Percent 0.4 Spouse 0.3 Head 0.2 0.1 Relative 0

Non-Kin

5

-

15 65 10 20 25 30 35 40 45 50 55 60 70

0

------

70+

5

10 60 15 20 25 30 35 40 45 50 55 65 Age Group

Figure 2-32. Percentage of sample by age and relationship to the household head.

Females, 1891 1 0.9 0.8 0.7 0.6 Children 0.5

Percent 0.4 Spouse 0.3 Head 0.2 0.1 Relative 0

Non-Kin

5

-

10

15 65 20 25 30 35 40 45 50 55 60 70

0 -

------

70+

5

10 60 15 20 25 30 35 40 45 50 55 65 Age Group

89 Figure 2-33. Percentage of sample by age and relationship to the household head.

Males, 1901 1 0.9 0.8 0.7 0.6 Children 0.5

Percent 0.4 Spouse 0.3 Head 0.2 0.1 Relative 0

Non-Kin

5

-

15 65 10 20 25 30 35 40 45 50 55 60 70

0

------

70+

5

10 60 15 20 25 30 35 40 45 50 55 65 Age Group

Figure 2-34. Percentage of sample by age and relationship to the household head.

Females, 1901 1 0.9 0.8 0.7 0.6 Children 0.5

Percent 0.4 Spouse 0.3 Head 0.2 0.1 Relative 0

Non-Kin

5

-

10

15 65 20 25 30 35 40 45 50 55 60 70

0 -

------

70+

5

10 60 15 20 25 30 35 40 45 50 55 65 Age Group

90 Figure 2-35. Headship rates by age and sex of head and census year.

Headship Rates, 1851 1 0.9 0.8 0.7 0.6 Males 0.5 0.4 Females

Percent Head Percent 0.3 0.2 0.1 0

Age Group

Headship Rates, 1861 1 0.9 0.8 0.7 0.6 Males 0.5 0.4 Females

Percent Head Percent 0.3 0.2 0.1 0

Age Group

91

Headship Rates, 1871 1 0.9 0.8 0.7 0.6 Males 0.5 0.4 Females

Percent Head Percent 0.3 0.2 0.1 0

Age Group

Headship Rates, 1881 1 0.9 0.8 0.7 0.6 Males 0.5 Females 0.4

Percent Head Percent 0.3 0.2 0.1 0

Age Group

92

Headship Rates, 1891 1 0.9 0.8 0.7 0.6 Males 0.5 Females 0.4

Percent Head Percent 0.3 0.2 0.1 0

Age Group

Headship Rates, 1901 1 0.9 0.8 0.7 0.6 Males 0.5 0.4 Females

Percent Head Percent 0.3 0.2 0.1 0

Age Group

93 Figure 2-36. Percentage of sample living in simple households by age and census year.

Males in Simple Households 0.45 0.4 0.35 0.3 1851 0.25 1861 0.2 1871 0.15 1881 0.1 1891

Percent Living Living PercentHousehold in Type 0.05 1901 0

Age Group

Females in Simple Households 0.45 0.4 0.35 0.3 1851 0.25 1861 0.2 1871 0.15 1881 0.1 1891

Percent Living Living PercentHousehold in Type 0.05 1901 0

Age Group

94 Figure 2-37. Percentage of sample living in extended households by age and census year.

Males in Extended Households 0.35

0.3

0.25 1851 0.2 1861 0.15 1871 0.1 1881 1891 0.05 Percent Living Living PercentHousehold in Type 1901 0

Age Group

Females in Extended Households 0.35

0.3

0.25 1851 0.2 1861 0.15 1871 0.1 1881 1891 0.05 Percent Living Living PercentHousehold in Type 1901 0

Age Group

95 Figure 2-38. Percentage of sample living in compound households by age and census year.

Males in Compound Households 0.8

0.7

0.6

0.5 1851 0.4 1861

0.3 1871 1881 0.2 1891 0.1 Percent Living Living PercentHousehold in Type 1901 0

Age Group

Females in Compound Households 0.8

0.7

0.6

0.5 1851 0.4 1861

0.3 1871 1881 0.2 1891 0.1 Percent Living Living PercentHousehold in Type 1901 0

Age Group

96 Table 2-11. Mean household size by type and census year.

Compound Extended Simple 1851 11.93 6.64 5.33 1861 11.00 6.80 5.24 1871 12.39 6.51 4.82 1881 11.23 6.55 4.82 1891 11.43 6.09 4.62 1901 11.08 5.80 4.33

Table 2-12. Mean valuation by household type and census year.

Compound Extended Simple 1851 22.98 16.49 13.39 1861 22.46 13.40 11.70 1871 28.98 14.46 10.16 1881 31.70 18.21 15.12 1891 34.03 9.52 12.79 1901 30.75 10.78 11.15

Table 2-13. Median valuation by household type and census year.

Compound Extended Simple 1851 9.82 6.83 5.63 1861 10.00 7.05 6.00 1871 10.00 7.20 6.00 1881 11.50 7.75 7.00 1891 10.13 7.00 6.00 1901 10.54 5.30 5.00

97 Figure 2-39. Histogram of valuation, 1881. 1881 is chosen as a reference year because it represents the maximum in median valuation. Valuation is measured in pounds. N=529, which is less than the total number of sampled households in 1881 (684) because of incomplete linkage of valuation and census records. This missing data may add bias to this socioeconomic

measurement.

.06

.04

Density

.02 0 0 100 200 300 400 500 Valuation

98 Table 2-14. Composition of households in 1851 by whether compound or not using a modified Hammel-Laslett classification. Compound household segments are used. The left columns add compound segments with non-compound segments for a hypothetical comparison.

99 Table 2-15. Composition of households in 1861 by whether compound or not using a modified Hammel-Laslett classification. Compound household segments are used. The left columns add compound segments with non-compound segments for a hypothetical comparison.

100 Table 2-16. Composition of households in 1871 by whether compound or not using a modified Hammel-Laslett classification. Compound household segments are used. The left columns add compound segments with non-compound segments for a hypothetical comparison.

101 Table 2-17. Composition of households in 1881 by whether compound or not using a modified Hammel-Laslett classification. Compound household segments are used. The left columns add compound segments with non-compound segments for a hypothetical comparison.

102 Table 2-18. Composition of households in 1891 by whether compound or not using a modified Hammel-Laslett classification. Compound household segments are used. The left columns add compound segments with non-compound segments for a hypothetical comparison.

103 Table 2-19. Composition of households in 1901 by whether compound or not using a modified Hammel-Laslett classification. Compound household segments are used. The left columns add compound segments with non-compound segments for a hypothetical comparison.

104 Table 2-20. Comparative proportions of family types. The family types in this table are defined using the same criteria as in Tables 23-28, but categories have been condensed (they would equal the ―total‖ rows in Tables 23-28. Sources for comparative data: a (Wall 2001:231); b (Laslett 1983:518-519); c (Laslett 1983:520-521); d (Laslett 1983:522-523); e (Laslett 1977:85); f (Laslett 1983:524).

105

Table 2-21. Percentages of extended, multiple, and complex households, Orkney and other European samples. Source for European data: (Wall 2001:222-223).

Country District Date Extended Multiple Complex N Scotland Orkney 1851 12 48 60 598 Orkney 1861 14 47 61 625 Orkney 1871 15 44 59 649 Orkney 1881 16 45 61 684 Orkney 1891 14 41 55 650 Orkney 1901 16 37 53 647 England Berks 1851 13 3 16 190 Buckingham 1851 12 2 14 413 Derby 1851 15 7 21 212 Derby 1851 16 7 23 154 Dorset 1851 12 2 14 407 Dorset 1851 11 1 12 264 Essex 1851 15 2 16 371 Gloucester 1851 14 3 17 95 Hertford 1851 14 6 20 326 Westmorland 1851 16 4 20 491 France Brittany 1851 12 8 20 197 Perigord Nord 1836 14 1 15 81 Corsica 1846 2 1 7 310 Corsica 1846 17 2 23 201 Corsica 1846 5 3 24 78 Pyrenees 1846 23 14 37 -- Germany Osnabrück 1858 -- -- 36 527 Switzerland Valais 1850 18 8 26 98 Iceland Arnesyssla 1845 -- -- 34 38 Gullbringusyssla 1845 -- -- 9 140 Sweden Uppland 1851 21 18 38 393 Västernorrland 1860 12 3 15 232 Blekinge 1850 2 1 3 115 Västergötland 1850 -- -- 11 287 Finland Åland 1840 18 15 34 65 Finland Proper 1859 16 7 23 728 Finland Proper 1859 19 7 26 106 Åland 1859 26 24 50 290 Osthrobothnia 1860 31 20 51 75 Viborg 1860 33 14 47 93 Portugal Minho 1850 12 14 27 210 Spain Basque 1842 -- -- 28 332 Castile 1860 4 1 5 3231 Murcia 1850 8 2 10 357

106 Table 2-21. Continued

Italy Piedmonte 1848 16 12 28 188 Piedmonte 1843 12 24 36 -- Piedmonte 1840 18 17 35 -- Friuli 1844 17 30 47 -- Friuli 1844 15 21 35 -- Friuli 1850 15 29 44 242 Emilia 1850 21 13 33 141 Emilia 1851 9 5 14 1523 Captianata 1838 14 9 24 1751 Bari 1839 9 2 11 1896 Bari 1855 10 6 16 1106 Campana 1856 4 4 8 2585 Estonia Southern 1850 17 31 48 451 Northern 1850 23 33 56 388 Saaremaa 1850 19 47 66 165 Lithuania Kurland 1858 24 24 48 92 Croatia Slavonia 1854 11 34 45 194 Bulgaria Southern 1838 20 50 70 44 Russia Ryazan 1850 7 66 73 166

107 References

Anderson, M. 1971. Family Structure in Nineteenth Century Lancashire. Cambridge: Cambridge University Press.

Andorka, R. and S. Balazs-Kovacs. 1986. "The social demography of Hungarian villages in the eighteenth and nineteenth centuries (with special attention to Sarpilis, 1792-1804)." Journal of Family History 11(2):169-192.

Bengtsson, T., C. Campbell, and J.Z. Lee. 2004. Life Under Pressure: Mortality and Living Standards in Europe and Asia, 1700-1900. Cambridge: MIT.

Bengtsson, T. and G.P. Mineau. 2008. Kinship and Demographic Behavior in the Past. New York: Springer.

Bengtsson, T. and O. Saito. 2003. Population and Economy: From Hunger to Modern Economic Growth. Oxford: Oxford University Press.

Berkner, L.K. 1972. "The stem family and the developmental cycle of the peasant household: An eighteenth-century Austrian example." The American Historical Review 77(2):398.

—. 1975. "The use and misuse of census data for the historical analysis of family structure." Journal of Interdisciplinary History 5(4):721-738.

Czap, P. 1982. "The perennial multiple family household, Mishino, Russia 1782-1858." Journal of Family History 7(1):5-26.

—. 1983. "'A large family: the peasant's greatest wealth': serf households in Mishino, Russia, 1814-1858." Pp. 105-152 in Family Forms in Historic Europe, edited by R. Wall, J. Robin, and P. Laslett. Cambridge: Cambridge University Press.

Dunbar, J.G. 1966. The Historic Architecture of Scotland. London: Batsford.

Engelen, T. and A.P. Wolf. 2005. Marriage and the Family in Eurasia: Perspectives on the Hajnal Hypothesis. Amsterdam: Askant.

Fenton, A. 1997. The Northern Isles: Orkney and Shetland. East Linton: Tuckwell Press.

Fenton, A. and B. Walker. 1981. The Rural Architecture of Scotland. Edinburgh: John Donald.

Goody, J. 1958. The Developmental Cycle in Domestic Groups. Cambridge: Cambridge University Press.

—. 1972. "The evolution of the family." Pp. 103-124 in Household and Family in Past Time, edited by P. Laslett and R. Wall. Cambridge: Cambridge University Press.

108 Hajnal, J. 1965. "European marriage patterns in perspective." Pp. 101-143 in Population in History, edited by D.V. Glass and D.E.C. Eversley. London: Arnold.

—. 1982. "Two kinds of preindustrial household formation systems." Population and Development Review 8(3):449-494.

Hammel, E.A. 1972. "The zadruga as process." Pp. 335-373 in Household and Family in Past Time, edited by P. Laslett and R. Wall. Cambridge: Cambridge University Press.

—. 1984. "On the *** of studying household form and function." Pp. 29-43 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

—. 2005. "Chayanov revisited: A model for the economics of complex kin units." Proceedings of the National Academy of Sciences 102(19):7043-7046.

Hammel, E.A. and P. Laslett. 1974. "Comparing household structure over time and between cultures." Comparative Studies in Society and History 16(1):73-109.

Janssens, A. 2002. Family and Social Change: The Household as a Process in an Industrializing Community. Cambridge: Cambridge University Press.

Jennings, J.A. 2010. "Household level predictors of the presence of life-cycle servants in northern Orkney, Scotland, 1851-1901." Presented at the annual meeting of the Population Association of America, Dallas, TX.

Jennings, J.A., C.S. Sparks, J.W. Wood, P.L. Johnson, T.M. Murtha, and S.A. Matthews. 2009. "Formation and dissolution of extended-family households in Orkney, Scotland, 1851- 1901." Presented at the meeting of the International Union for the Scientific Study of Population, Marrakech, Morocco.

Kahk, J., H. Palli, and H. Uibu. 1982. "Peasant family and household in Estonia in the eighteenth and the first half of the nineteenth centuries." Journal of Family History 7(1):76-88.

Kertzer, D.I. 1984. Family Life in Central Italy, 1880-1910: Sharecropping, Wage Labor, and Coresidence. New Brunswick: Rutgers University Press.

Kertzer, D.I. and M. Barbagli. 2002. The History of the European Family: Family life in the long nineteenth century (1789-1913). New Haven: Yale University Press.

Laslett, P. 1977. "Introduction." Pp. 23-89 in Household and Family in Past Time, edited by P. Laslett and R. Wall. Cambridge: Cambridge Univ. Press.

—. 1983. "Family and household as work group and kin group: areas of traditional Europe compared." Pp. 513-564 in Family Forms in Historic Europe, edited by R. Wall, J. Robin, and P. Laslett. Cambridge: Cambridge University Press.

—. 2000. The World We Have Lost: Further Explored. London: Routledge.

109 Laslett, P., J. Robin, and R. Wall. 1983. Family Forms in Historic Europe. Cambridge: Cambridge University Press.

Laslett, P. and R. Wall. 1974. Household and Family in Past Time. Cambridge: Cambridge University Press.

Mitterauer, M. 1992. "Peasant and non-peasant forms in relation to the physical environment and the local economy." Journal of Family History 17(2):139-159.

Moring, B. 1999. "Land, labor, and love: Household arrangements in nineteenth century eastern Finland--cultural heritage or socio-economic structure?" The History of the Family 4(2):159-184.

Murtha, T.M., P.L. Johnson, J.W. Wood, S.A. Matthews, C.S. Sparks, and J.A. Jennings. 2008. "Historical demography, oral history, settlement archaeology, and landscape ecology: The North Orkney Population History Project." Presented at the annual meeting of the European Population Conference, Barcelona, Spain.

Naismith, R.J. 1985. Buildings of the Scottish Countryside. London: Victor Gollancz in association with Peter Crawley.

Netting, R.M. 1993. Smallholders, Householders: Farm Families and the Ecology of Intensive, Sustainable Agriculture. Stanford: Stanford University Press.

Neven, M. and C. Capron. 2000. Family Structures, Demography and Population: A Comparison of Societies in Asia and Europe. Liege: University of Liege.

Palsson, H. and P. Edwards. 1981. Orkneyinga Saga: The History of the Earls of Orkney. London: Penguin Books.

Reay, B. 1996. Microhistories: Demography, Society and Culture in Rural England, 1800-1930. Cambridge: Cambridge University Press.

Ruggles, S. 2009. "Reconsidering the Northwest European family system: Living arrangements of the aged in comparative historical perspective." Population and Development Review 35(2):249-273.

Segalen, M. 1983. Love and Power in the Peasant Family. Rural France in the Nineteenth Century. Oxford: Blackwell.

—. 1984. "Nuclear is not independent: Organization of the household in the Pays Bigouden Sud in the nineteenth and twentieth centuries." Pp. 163-186 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

Schrank, G. and D.G. Lockhart. 1995. An Orkney Estate: Improvements at Graemeshall, 1827- 1888. East Linton: Tuckwell Press.

110 Sparks, C.S. 2007. Households, Land, and Labor: Population Dynamics in the Northern Orkney Islands, Scotland, 1851 to 2003. Ph.D. Dissertation, Anthropology and Population Research Institute, Pennsylvania State University.

—. 2009. "An application of the variable-r method to subpopulation growth in a 19th century agricultural population." Demographic Research 21:23-64.

Szoltysek, M. 2007. "Central European household and family systems, and the ‗Hajnal– Mitterauer‘ line: The parish of Bujakow (18th–19th centuries)." The History of the Family 12(1):19-42.

—. 2008. "Three kinds of preindustrial household formation system in historical Eastern Europe: A challenge to spatial patterns of the European family." The History of the Family 13(3):223-257.

Thomson, W.P.L. 1983. Kelp-making in Orkney. Stromness: The Orkney Press.

—. 2008a. The New History of Orkney. Edinburgh: Birlinn Ltd.

—. 2008b. Orkney Land and People. Kirkwall: The Orcadian Limited (Kirkwall Press).

Wall, R. 2001. "The transformation of the European family across the centuries." Pp. 217-241 in Family History Revisited: Comparative Perspectives, edited by R. Wall, T.K. Hareven, and J. Ehmer. Newark: University of Delaware Press.

Westray Buildings Preservation Trust. 2002. Fae Quoy tae Castle: An Orkney Island's Snapshot in Time. Kirkwall: Orcadian Limited.

Whyte, I.D. 2007. "Cumbrian village communities: Continuity and change, c. 1750-1850." Pp. 96-113 in The Self-Contained Village? The Social History of Rural Communities, 1250- 1900, edited by C. Dyer. Hertfordshire: University of Hertfordshire Press.

Wilk, R.R. 1984. "Households in process: Agricultural change and domestic transformation among the Kekchi Maya of Belize." Pp. 217-244 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

—. 1991. Household Ecology: Economic Change and Domestic Life Among the Kekchi Maya in Belize. Tucson: University of Arizona Press

Yanagisako, S.J. 1979. "Family and household: The analysis of domestic groups." Annual Review of Anthropology 8(1):161-205.

Chapter 3

Dissolution and Formation of Extended-Family and Compound Households in Northern Orkney, Scotland, 1851-1901

Introduction

Historical demographers have long debated the nature of households in past societies, with much attention paid to the categorization of household types and the analysis of their geographical distributions and changes in prevalence over time (Berkner 1972; Hajnal 1982;

Hammel and Laslett 1974; Kertzer 1989; Laslett and Wall 1977; Wall 1983). In the traditional farming system that dominated the economy of preindustrial Orkney in Scotland, households provided the labor needed to produce food and other goods for family consumption. In order to improve their odds of a successful subsistence enterprise, these smallholder households needed to balance the group‘s current age and sex composition with both current and future food consumption and labor requirements (Chayanov 1986; Hammel 2005; Netting 1993). Household composition can be considered an important factor in determining a household‘s energy requirements as well as its ability to muster enough labor to produce the goods necessary to satisfy those needs. When considered in conjunction with agricultural resources, such as arable land, to which the household has access, a household‘s composition may be an important determinant of the economic and physical well-being of its members, especially at times of stress such as might be expected during food shortages or the illness or death of household members.

This study examines households as dynamic entities that are at risk of transition between household types. Specifically, the expansion of simple households to extended or compound households and the breakdown of extended and compound households into simple ones are

112 modeled and described. These transitions may be a response to economic conditions (either internal or external) of any given household, demographic fortunes (including births and deaths), the productive resources to which a household has access, or a combination of these factors.

Household transitions are modeled using discrete-time event-history models. These models allow for the use of longitudinal data that are interval censored, and time variant and time invariant covariates can be included. In conjunction with descriptive statistics, extended, compound, and simple households are compared and some potential risk factors for transitions between household forms are identified.

Background

Setting and Data Sources

The Orkney Islands are an archipelago off the coast of northern Scotland, where the

North Sea and Atlantic Ocean meet. The six islands of the study are Westray, Sanday, Papa

Westray, Eday, North Ronaldsay, and Faray. These islands were selected as part of an ongoing interdisciplinary study of population and family history, settlement, and land use known as the

North Orkney Population History Project (Jennings et al. 2009; Murtha et al. 2008; Sparks 2007,

2009). The northern Orkney Islands were, and have remained, rural in character. Historically, agricultural production relied on grains, mainly black oats and bere (a landrace of barley), and traditional breeds of livestock, including sheep, cattle, chickens, and pigs. These activities supported household subsistence and paid rents. Individual farmsteads, or groupings of houses and agricultural buildings associated with gardens and fields, are dispersed over the islands with a few small villages that are slightly more densely settled, but still mainly agricultural in nature.

These farmsteads all have names and these names appear in historical records and persist for

113 many generations, even while the inhabitants change over time (Palsson and Edwards 1981;

Thomson 2008a, 2008b). Cadastral (estate) maps from the decades before the study period, such as the 1830s and 1840s, feature farmsteads with the same names and locations as in later historical and modern maps. Orkney historians speculate that most of the farm and house names became fixed by the 1840s to accommodate record keeping (such as census enumerations and valuation rolls) and mail delivery (Thomson 2008a). The population of the islands reached its height in the mid- to late-1800s, after which out-migration, both to mainland Scotland and overseas, caused the depopulation that continues to this day. Depopulation and the widespread use of stone as a building material have facilitated the survival of archaeological remains of farmsteads. The remains allow for the study of the settlement history of the islands and provide information about the structures present at farmsteads that appear in historical records.

Farmstead names are listed in census returns, along with the inhabitants of the farmstead and their relationship to the household head. This allows for the tracking of household size, composition, and type over time. Farmsteads are also listed in vital records, so that births and deaths can be linked to the household in which they occurred. Valuation rolls, which are records of property value on which taxes were assessed, are also linked to census returns using farm names. These records are a good proxy measure for landholding size and quality, as well as rental value.12 These three sources allow for longitudinal study of households, including their transitions between household types, as well as the observation of household-level characteristics that may be predictive of these transitions.

12 See Appendix A for details about how this assessment was made.

114 Households as Dynamic Processes

Social scientists had long considered the stem family the predominant family form in preindustrial Europe. Based largely on the work of Le Play (see 1982, for a collection of translated works), this once popular idea was refuted by Laslett and his colleagues at the

Cambridge Group for the History of Population and Social Structure (Laslett 1984b; Laslett et al.

1983; Laslett and Wall 1974). Using census records and census-like listings of English communities from1574-1821, this group of scholars refuted the popular notion of the prevalence of large, extended families and asserted that since the medieval period, European families were mostly small and nuclear in structure. As Peter Laslett (1984:90-91) famously put it,

[It is commonly held] that our ancestors lived in large familial units. Family groups, it seems to be almost universally agreed, ordinarily consisted in the pre- industrial past of grandparents, children, married as well as unmarried, grandchildren and often relatives, all sleeping together in the same house, eating together and working together.... [If so,] households would have had to be bigger than our households are, and more complicated in their inner relationships as well: extended families is the phrase which is almost always used. Now all these statements have been demonstrated to be false.... It is not true that most of our ancestors lived in extended families (Laslett 1984b).

This demographic approach to family history revolutionized the field, but it was also the subject of subsequent debate and criticism.

Two important weaknesses in the Cambridge Group‘s work were identified by critics

(See Anderson 1995 for a review of other criticisms). The first relates to statements about the generalization of English data to the whole of northwest Europe. Studies of areas outside of

England reveal greater diversity in family forms than originally asserted by the Cambridge Group

(Berkner 1972; Czap 1982; Goubert and Denault 1977; Kertzer 1984; Plakans and Wetherell

2005; Szoltysek 2007; Viazzo 2005). Evidence presented here from Orkney, Scotland, follows in this tradition of local area studies that demonstrate variations on the broad family patterns

115 described by the Cambridge group. This new information on a specific location suggests the need to modify generalized statements about the incidence of nuclear and non-nuclear family forms.

A second weakness of the Cambridge Group‘s approach involves their use of cross- sectional data and systems of household classification. In their studies of English communities, households were identified in lists of community members and then classified using a typology based on the relatedness of individuals to the household head (Hammel and Laslett 1974; Laslett

1977). From this information, the percentage of nuclear and extended households and the average size of households were obtained. Critics noted that these cross-sectional measures could not address changes in households, thus transforming a dynamic process of household development into a static typology. Family types or categories, in this view, should not be confused with family systems, which characterize family processes, family cycles, and family development (Skinner 1997). Several scholars note that stem-family systems, in which an heir

(and his/her spouse and children) remain in the parental home, will appear to be predominantly nuclear when applying the cross-sectional classification method (Berkner 1972, 1975; Flandrin

1976). This is because over the course of its existence, a stem household will alternate from nuclear to extended types several times, so that cross-sectional measurements of household types will underestimate the percent of households that ever pass through a phase of household extension. Hammel has characterized this as a problem of the misinterpretation of ―snapshots‖ of behavior (1972). It is also important to consider that the household classification method lacks the context needed to distinguish households of similar forms that may have been produced by different means, and therefore have different meanings in terms of the development and trajectory of that particular domestic group (Kunstadter 1984; Segalen 1977).

Household cycles have been described in terms of stages. These might include expansion, dispersion or fission, and replacement (Fortes 1958). Other conceptions include transitions between household types, such as nuclear, stem (1 child married), joint (2 children

116 married), and division (Wakefield 1998). Based on similar ideas of phases in the household cycle the comparative study of household types has been attempted using a variety of methods. For instance, some have tracked family types by the age of the head (or other focal household member) to describe how household composition changes over an individual member‘s life course (Anderson 1971; Katz 1975; Sieder and Mitterauer 1983; Wolf 1984).

Chayanov proposed a family life-cycle schema that considered the economic effects of internal changes in household composition (1986). He noted that the balance of producers to consumers changes over the family life-cycle. He tracked the establishment of a household by a newlywed couple and observed that as children are born the ratio of consumers to producers (C/P ratio), calculated using a set of age- and sex-based weights to signify individual contributions to production and consumption needs, becomes unfavorable and remains so until children become old enough to contribute to productive work. In this view of the family life-cycle, the age and sex composition of the family determines its productive potential as well as its consumption requirements. Demographic circumstances, therefore, produce variation in the economic well- being of households such that both family- and individual-level changes contribute to cycles in household C/P ratios. This summary measure of household composition may be an important predictor of transitions between household forms.

However, it should be noted that these broad generalizations about household cycles can be misleading, as all households do not pass through all phases, and households do not experience phases in ―lock-step‖ (Mitterauer and Sieder 1979). Others have noted that these life-cycle analyses do not adequately address the dynamic process they claim to study; rather, they replace one ―snapshot‖ with a series of snapshots (Alter 1988; Elder 1978). This problem results from the inherent difficultly in drawing inferences about a cyclical process from cross-sectional or semi-longitudinal data (Katz 1975). In addition, a number of other processes add complexity to the analysis of household cycles. Individual, family, and household life-cycles exist

117 simultaneously and are likely to affect one another (Berkner 1972; Chayanov 1986; Mitterauer and Sieder 1979). It is also important to distinguish cyclical and secular processes (Hammel

1972), as changes that occur as part of the family cycle may be confused or confounded with changes that occur over time as population-level demographic trends shift.

In response to these shortcomings of family life-cycle studies, other strategies have been developed to address the problem of household dynamics. Some have modeled household histories as Markov processes (Carter 1984; Fjellman 1977; Otterbein and Otterbein 1977).

However, Markov models may not be a good choice, as household histories are not a series of states that exist without a ―memory‖ of past events (Carter 1984). Others have used continuous population registers to track households and their members through time (Alter 1988; Janssens

2002). Unfortunately, the data from Orkney will not support this type of analysis. Van de Walle has approached the study of household dynamics by focusing on the transitions between household types (1976). This method avoids many of the difficulties associated with family cycle studies mentioned above, as it does not assume an order or age-progression of states. The current study takes this household type transition method a step further by introducing an event-history model of transitions between household types. Event-history models allow for the consideration of the influence of both constant and time-varying covariates on the likelihood of a transition.

Households in Orkney: Simple, Extended, and Compound Types

For the purposes of this study, household types are defined following the classification system devised by Hammel and Laslett, with one notable exception (1974). Simple households consist of a married couple, or a surviving member of that couple, and their children.13 Extended

13 These definitions refer to related individuals within the household. The presence of servants, lodgers, boarders, and visitors is not inconsistent with simple households as defined here. Thus, a household that contains a married couple, their children, a servant, and a lodger is classified as a simple household.

118 households, in contrast, feature one or more related individuals additional to the simple household. In this study, ―extended households‖ combines two of Hammel and Laslett‘s categories, extended households and multiple-family households, which include two or more married couples (Hammel and Laslett 1974). The surviving archaeological remains in Orkney provide evidence for a third household type, the compound household, not included in the

Hammel-Laslett classification, which would otherwise go undetected. While demographic data sources suggest that extended households are less common than simple households, the physical remains of household-based farmsteads reveal that many units listed separately in the census were actually single households in the economic sense described by Hammel (2005). These compound households, often linked by brothers, are adjacent, or even structurally joined, and share a common set of essential farming structures such as barns, byres, grain kilns, and stables. Given what has been learned about Orcadian economic and social systems through this interdisciplinary project, the investigators are confident that the component units of such compound households were not independent, but rather cooperatively worked their holdings and shared the products of their labor, even if some members were employed in outside wage labor, which was often part- time or seasonal. By this criterion, the sharing of a single set of farming structures, compound households comprise 26% to 41% of the households in the sample, whereas extended households represent between 18% and 25% of sampled households (Table 3-1). Compound households, particularly when compared with extended households, provide a good test case for predictions about the economic impetus for forming non-simple households and the ways in which landholding size and quality may place an upper limit on household size and degree of extension.

A particular case will serve as illustration. On the island of Westray, a croft of 1.8 hectares called South Hammer was abandoned in the 1980s and absorbed into a neighboring farm. According to the 1901 UK census, the inhabitants of South Hammer were as listed in Table

3-2. If that record were the only source of information available, would we code South Hammer

119 as a single extended family or as three more or less independent nuclear-family households?

Since a ―head‖ is listed for each of the three units, one might be inclined to call them separate households, especially if we have in mind Laslett‘s dictum about extended families being anomalous in this part of the world. Archaeological investigation and detailed mapping, however, suggest we would be wrong. The project‘s survey of the physical remains of South

Hammer show that the three domiciles were either very close or attached; more importantly, it shows that the entire complex had but one barn, byre, stackyard, kailyard, and muckyard, the minimal structures needed for a single farming unit (Figure 3-2). Moreover, the 1901 OSGB map

(Figure 3-3) shows that the same structures existed contemporaneously with the inhabitants in

Table 3-2. Whatever else they may do, including part-time and seasonal work on other farms or as fishers or craftsmen, crofters are farmers on their own holdings and everyone living at the croft helps with the farming if physically able. For example, several occupants of South Hammer have listed occupations other than farmer. In the case of two of these individuals, identified as ploughmen at Tirlot (a large nearby farm), this work was seasonal, as plowing only occurs during certain weeks of the year, but was probably specified as their ―occupation‖ because it provided wages, unlike their work at home. The same is probably true for the household members listed as fishermen. Therefore, South Hammer should be regarded as one extended household whose members pool resources as a single farming unit.14

This type of compound household is not limited to Orkney. Compound households, in which dwelling units are contained within an outer wall and whose members share production activities and commonly pool resources, are found throughout West Africa (Goody 1958). Wilk describes ―household clusters‖, or households composed of more than one dwelling unit that share economic activities in Belize (1984; 1991). Arrangements similar to those described at

14 Genealogical linkages reveal that the three ―heads‖ in Table 3-1 are all full siblings who grew up at South Hammer (Figure 3-1).

120 South Hammer have been documented in 19th century France (Segalen 1983, 1984). In addition,

Berkner disagrees with some of Laslett‘s definitions, arguing that in Austria, retired parents living in separate quarters or a cottage on the same premises as their children (as is often the case in

Orkney as well) should be considered a single household rather than two independent households

(1975). He goes further to suggest that the interpretation of household divisions and boundaries is a difficult task that should be based on knowledge of local contexts (Berkner 1975). In later work, Laslett recognizes the inability of current historical demographic methods to identify the equivalent of compound households. He writes:

Some lists of inhabitants do, like the Census of 1851 [England], mark houseful divisions as well as household divisions, but none specify the kin relationships within housefuls [see below]. Unfortunately the present techniques of historical demography cannot be used to determine any but a few such relationships existing in any community. Hence, although the shape of housefuls—when and where distinct from households—is so rarely indicated there is always a possibility that these larger units themselves constituted some sort of extended family associations (1977:36-37).

Fortunately, archaeological and documentary information from Orkney allow for the identification of compound households (or housefuls15 with kin relationships in Laslett‘s terminology) and in many cases, the reconstruction of household histories and kinship links within the household as well.

By this criterion, the sharing of a common set of farming structures, household extension is, in accord with Laslett, not ubiquitous in North Orkney but nor is it terribly rare, and its frequency varies extensively across islands and periods (Table 3-1). Why is the frequency of extension in Orkney high when compared to the rest of northwest Europe (Chapter 2)? And, if household extension is advantageous under some circumstances, why is it not even more

15 Laslett‘s (1977) use of the term ―houseful‖ corresponds roughly to what is called a ―compound household‖ here. In his usage, households must occupy a single dwelling, but multiple dwellings may exist on a shared premises, such as a farmstead or apartment building. The combined membership of these multiple dwellings constitutes a houseful. I argue that compound households in Orkney should be considered households, rather than housefuls. In Orkney, kin relationships can be specified and economic interdependence can be inferred for compound households.

121 common? Some recent ideas from Gene Hammel may provide a way to think about these questions and incorporate them into event-history models of household transitions (2005).

Building upon models of the economics of the household life cycle originally developed by Chayanov (1986), Hammel suggests that household extension may dampen unfavorable fluctuations in household consumer/producer (C/P) ratios by combining nuclear-family units at differing phases of their life cycle. While this improvement in C/P ratios may be an impetus for forming extended-family households, in theory there is an important limit to the process that was ignored by Hammel: as households become more extended, they also, generally speaking, become larger. In rural Orkney, allotments of arable and pasture were fixed (at least over a time scale of the period under examination) and could not be expanded as the household grows. It is hypothesized, therefore, that household size should be an important predictor of the dissolution of extended households, at least when controlled for size of holding (both its main effect and in interaction with household size). In addition, C/P ratios will be included as time-varying covariates in event-history models of household transitions, as they provide information about household composition with respect to the production potential and consumption requirements of the household. Information from prior historical and ethnographic studies suggests other potential predictors of household transitions and circumstances under which smaller simple households or larger extended and compound households would be preferable.

Possible Determinants of Household Transitions

If they have sufficient land, large households have certain economic advantages over smaller households. Extended-family living arrangements, which on average include more people than simple-family households, may be economically advantageous as they allow for a potentially larger and more diverse labor pool. In addition, extended households may have served

122 several other important functions in this rural island community, including shared farm labor among related households, childcare, and assistance in establishing migration networks.

Extended households may have also functioned as social safety nets by which care for dependents, such as children, the elderly, and the infirm were provided (Das Gupta 1997a; Goody

1996; Van de Walle 1976; Wall 1983). Many extended households are classified as such because of the presence of one or more elderly relative, and the frequency of such three-generation households is often linked to prevailing mortality rates and mean age at marriage (Burch 1970;

Coale 1965; Hajnal 1965). It may be the case that these types of households are less stable than larger extended-family households, as the loss of the one ―extending‖ member would cause a change in household form, so household size will be included in the event-history models as a time-varying covariate to control for the effects of overall household size. With respect to infants, 19th century Belgian multiple-family households have been observed to have an abundance of children and young couples (Van de Walle 1976). However, in cases where there are dependent household members (either young or old), these additional individuals may actually represent a net drain on the household‘s resource base rather than a net gain, making the ratio of producers to consumers less favorable. Moreover, the potential gains of larger, extended families must also be balanced against other concerns, such as overcrowding within dwellings and declining marginal returns on additional labor (Chayanov 1986; Das Gupta 1997b; Hagen et al. 2006). Indeed, in examples of extreme poverty, households have been observed to split apart

(Boulton 2000). It is therefore necessary to consider extended-family living arrangements dynamically, with particular attention to changes in household composition, access to land, and land ownership. The study of the dynamic processes of extended household formation and dissolution can therefore provide insight into the potential advantages and disadvantages of these living arrangements.

123 Social scientists have considered a variety of potential reasons for dividing large households. These include the reduction of tax burdens (if taxes are assessed on the size of the household or its dwelling), disagreements among kin, attempts to escape poverty, overcrowding, and individual motivations involving the division of household resources among co-resident nuclear families (Wakefield 1998). Indeed, these internal conflicts represent what Laslett called the ―knot of individual interests‖ that comprises households (Laslett 1984a). Many authors have cited conflicting strategies within households as a cause for the division of large households into two or more smaller ones (Das Gupta 1999; Fontaine and Schlumbohm 2000; Hammel 2005;

Netting 1993).

Smaller households have, on average, smaller kinship networks and fewer sources of help to call upon in times of need than larger households (Laslett 1988). Smaller households may wish to expand their membership to meet economic requirements. Sharecropping households in

Italy, for example, tend to be larger and more complex than day-laboring households (Kertzer

1984; Viazzo 2005). In these households, household size is correlated with the size of landholdings. Several studies of traditional farming societies have found similar associations

(Erdozáin-Azpilicueta and Mikelarena-Peña 1998; Netting 1993; Rudolph 1992; Viazzo 2005;

Viazzo et al. 2005). Larger households tend to have (or are able to gain) access to more land.

Information from the valuation rolls provides a good proxy for the amount of land to which a household has access. This variable will be modeled as a time varying covariate in the event- history models of household transitions. Larger households have also been associated with landownership or tenancy rather than day-laboring (landless) status (Alderson and Sanderson

1991; Segalen 1977). Information about the landowning status of household heads is available from the valuation rolls and will be included in the event-history models as a time varying covariate. High workloads and a mixed farming economy have also been linked to larger households (Anderson 1995; Gaunt 1978). Many have suggested that nuclear households change

124 to extended households as part of an economic strategy to obtain sufficient labor or increase production (Czap 1983; Reyna 1976; Rudolph 1992; Wilk and Netting 1984). However, the ultimate economic cause of household extension may vary by context, as there is disagreement in the literature about the economic causes of the transition to extended-family forms. Some studies have noted that families struggling to make ends meet keep producers, especially grown children, in the household longer or reside together to split housing and other costs (Anderson 1971;

Spagnoli 1983). Others propose that adding kin to the household is a luxury, and not a response to economic hardship (Ruggles 1987).

In addition to economic and land-related causes for changes in household composition, demographic processes contribute to transitions between household types. Anderson describes how Lancashire families changed their membership to adapt to ―critical life situations‖ that include births, deaths, illnesses, and spells of unemployment (1971). Others have noted the importance of births, deaths, marriages, and retirements to the development of the household cycle and changes in household forms over that cycle (Berkner 1972). Therefore, it is important to consider the potential influence of vital events on the risk of transition from one household form to another. In this study, the number of births and the number of deaths that occur in the household during the intercensal period will be included as covariates in the event-history models of the formation and dissolution of extended and compound households. Ideally, marriages should be included as well, but the current state of data linkage does not allow for the tracking of a married couple to their post-marital residence.

Data and Methods

This study combines historical documentary evidence and archaeological survey data.

Three primary historical data sources are used: vital registration records, individual-level census

125 returns, and valuation rolls. The historical data were collected from the General Registrar of

Scotland (GROS) and the Orkney Library and Archive and the archaeological surveys were conducted over several field seasons, each as part of an ongoing study of the population history of the northern Orkney Islands (Murtha et al. 2008; Sparks 2007). The decennial censuses provide information on every person at home on the day of enumeration including visitors and boarders (a de facto census), including age, sex, and marital status (Cory 2004; The National Archives of

Scotland 2003). These returns provide the composition of the household. Information about vital events that occur in each household is taken from vital records of births and deaths, which include the name of the farmstead or house in which the individual was present.16 These house names, which persist over time, are used to link the census returns to the vital records. Valuation rolls, or records of the taxation or rental value of land and buildings, also list house names, which are again used to link these data to information from the censuses and vital registers. The data used in this study represent all households on the islands of Westray, Eday, Papa Westray, and Faray for which such record linkage was possible.

The sample used in this study comes from households on the islands of Westray, Eday,

Papa Westray, North Ronaldsay, and Faray. Sanday was excluded from this sample because records of vital events that occurred on this island have not yet been linked to their location using house names, so the vital events cannot be linked to the census and valuation records. This form of record linkage may introduce bias into the sample, as only households whose names could be identified and tracked over time were selected into the sample. Therefore, households with house names that only existed for a few years are less likely to be linked across census years. Very small crofts and farmsteads that were abandoned early in the period of interest are more likely to

16 Marriage and migration are important demographic events that are not included in these analyses. Marriage records have not yet been linked using house names. Once this work is complete, models may include the number of marriages to men and women as covariates. Migration cannot be directly observed using the data available. This may be a source of bias in both the sample and the analyses.

126 be missing from historical maps and oral traditions regarding place names. Very large holdings and manses (houses assigned to the clergy) were excluded from the sample, as these households are presumably subject to different pressures than the average tenant farmer.17 For example, very large holdings were owned by absentee landlords and managed locally by factors. Day-laborers were often hired to augment the labor of residents who organized agricultural production. Clergy were given housing and stipends and did not practice subsistence farming (although they did participate in agriculture as a side-employment). This sample selection criterion focuses the sample on subsistence farmers. Later analyses (or re-analyses) will include the island of Sanday once data linkage is complete and compare the dynamics of smallholding households to large- scale farmers and other community members, such as the clergy. Because the population of the islands changes over time, the precise sample size changes by census year (Table 3-1).

Households were observed at each decennial census from 1851 to 1901. Information about household composition and type is known only from census data so that the events of interest, the formation and dissolution of extended and compound households, are interval censored. Because of interval censoring, discrete-time logit hazard models were used to predict the dissolution or formation of households. For the purposes of this study, extended households are identified using the criteria outlined in Hammel and Laslett (1974). Households with more than one head listed in the census, are categorized as compound households. Simple households are at risk of forming extended or compound households, so they are included in the formation models, but not the dissolution models, because they are not in that risk set. The estimated hazard function of the break-up of extended households decreases over the study period, while the hazard of the break-up of compound households increases (Figure 3-4). The hazard of the

17The specific properties excluded from the sample include Brough, Broughton, Cleat, Links, Pierowall, Gill Pier, Trenabie, Tuquoy, Clifton, Skaill, Noltland, and the Manses on the island of Westray; Holland and the Manse on Papa Westray; Carrick, Greentoft, and the Manses on Eday; and the Manse on North Ronaldsay. All of the properties on Faray were included in the sample.

127 formation of extended households increases over the study period, whereas the hazard of the formation of compound households decreases (Figure 3-5). At this time, it is speculated that the increasing prevalence of extended households and decreasing prevalence of compound households may be attributable to depopulation or changing economic conditions (see Chapter 1).

Although these hazard trends are not always monotonically increasing or decreasing, time was modeled using a linear function. The linear time function had better model fit when compared to a general (piecewise) time function, as measured by likelihood-ratio tests and the Bayesian

Information Criterion (BIC), and did not significantly affect parameter estimates and z-scores.

Several models of the effect of household size on the breakdown and formation of extended and compound households were estimated with controls for the linear effects of time, as measured by census intervals, and other covariates. Household size was modeled as a time- varying covariate with no lags, such that the relevant household size for a given dissolution or formation event was that immediately prior to the event. Births and deaths that occurred in the census interval were also modeled as time-varying covariates. While the exact dates of the births and deaths are known from vital registers, it was necessary to sum their respective numbers over the census interval to make them consistent with the intervals in which the outcome variable is observed. Land values and ownership status were modeled as time-varying covariates, measured at each census year, even though their values changed little over time. Consumer-producer ratios were computed for each household at each census interval using the weighting system outlined by

Hammel (Table 3-3). This weighting system was chosen because it included earlier productive contributions of children than was proposed by Chayanov. Given historical and oral accounts of farm life, it is reasonable to believe that children in 19th century Orkney began assisting on the farm, albeit in limited capacities, at young ages. Findings from a recent comparison of weighting systems used in the calculation of C/P ratios has found that the specific choice of weights used

128 has little effect on the direction and size of estimated coefficients when used as predictor variables (See Appendix B for details).

Results

Simple households range between 39 and 50 percent of the sample. Among simple- family households, the majority consist of a married couple with or without their children. The percentage of widow- or widower-headed households varies from 9 to 15 percent of all simple households (Table 3-4). Extended households are more likely to be extended upwards or downwards (as in stem families) than laterally. Multiple-family (but not compound) households, which have at least two married couples (or remaining widows/widowers of two couples) account for between 16% and 25% of all extended households (Table 3-5). The most common compound households consist of two nuclear family units (Table 3-6). However, few compound households contain a multiple-family component unit. One might hypothesize that compound households are a way for multiple-family households to maintain close economic ties while ensuring greater privacy in living arrangements, as was the case at South Hammer.

Simple, extended, and compound family households differ not only in their composition, but in other measures of interest as well. Compound households tend to include more people than simple or extended households, which is expected given the presence of additional dwelling space

(Table 3-7). Consistent with their larger size, compound households experience more vital events

(births and deaths) than extended or simple households (Table 3-8). Compound households also have higher tax valuations, which indicate that they have access to more land than smaller households (Table 3-9). The three household types have similar average C/P ratios, but these ratios vary more in simple households, as predicted by Hammel (2005) (Table 3-10). Finally, simple, extended, and compound households have different land ownership patterns, with more

129 simple and extended families owning their own land or renting from non-laird (lairds are wealthy land owners, usually absentee, with many holdings) owners than is expected, even though there are few owner-occupiers and non-laird landowners in this period (χ2 = 9.76, df = 4, p = 0.045,

Table 3-11). Although most of the differences among household forms in household size, valuation, number of vital events, C/P ratio, and landowner status are small, changes in these variables may be important in predicting transitions from one household form to another. Since many of these measures change over time, a variable that is a linear combination of ten-year census periods will be included in each model.

Dissolution of Extended and Compound Households

The results of the discrete time hazard models of household dissolution are given in Table

3-12. These models are stratified by the type of household (extended or compound) that dissolves. The main effects model and a model that includes the interaction of land value with household size are presented here. Models were estimated for other interactions of main effects, but only the interaction of land value and household size is reported here, because among interaction models, those that included this term had the best model fit, in terms of BIC (although not better than the main effects model), and the interaction term was sometimes significant.

For both household types, household size and C/P ratio are significant (or nearly significant) at the p = .05 level in the main effects model. The coefficients of these terms are similar between the two household types, with household size having a small, negative effect on the risk of dissolution, and C/P ratio having a larger, positive effect. However, extended and compound households differ with respect to the effects of births and deaths on the risk of household dissolution. In extended households, deaths significantly increase the rate of dissolution, whereas births have a negative but non-significant effect. In contrast, births have a

130 significant negative effect on the risk of dissolution of compound households, while deaths are not significant.

There are also differences between the two household types in the interaction models.

When the interaction of land value and household size is considered for compound households, the effect of household size becomes non-significant, while the coefficient for land value increases and approaches significance (p = .062). In the case of extended households, the inclusion of the interaction term slightly changes the effect of household size, which is no longer significant, but not land value. Thus, the initial counter-intuitive finding that large households are less likely to split than smaller households does not hold when considering the interaction of household size with the amount of land available. In both models, the coefficient of C/P ratio was positive, indicating that as the number of consumers relative to producers increases, or the economic situation of the household is less favorable, the more likely a household is to break apart. This finding is interesting and worthy of further investigation. If extended and compound households are, in general, economically advantageous to nuclear households, why are households with unfavorable C/P ratios more likely to split up? It may be the case that individuals in struggling households often leave to seek work or other economic opportunities outside of the household, thereby contributing to the risk of the household returning to a nuclear form.

The differential effect of births and deaths on the risk of household dissolution in the two household types is another interesting finding. One might expect that the number of deaths is more likely to contribute to the dissolution of extended households, which on average are smaller than compound households. However, potential explanations for the differential effect of the number of births on the dissolution of compound households are likely to be more complicated and less intuitive. For both household types, a greater number of births in a household reduces the risk of dissolution, but the term is only significant in compound households. The potential

131 implications of this finding will be taken up below in the context of the results of the formation models.

Formation of Extended and Compound Households

The results of the discrete time hazard model of household formation, stratified by household type, are given in Table 3-13. Again, both the main effects model and a model that includes the interaction of land value with household size are presented. As in the case of household dissolution, the interaction models were estimated for other interactions, but only the interaction of land value and household size is included because these models have better fit in terms of BIC, although not better than the main effects model. The results of the models of household formation for each household type vary more than they do for household formation, so the two types will be discussed separately.

In the main effects model of compound households, unfavorable C/P ratios reduce the risk of formation and the effect size is large (β = -1.54). If it is true that one step in the formation of a compound household is the addition or renovation of a dwelling structure at an existing farmstead, then it is reasonable that families that are struggling economically may not have access to the resources required to construct this defining feature of a compound household. Births significantly increase the risk that a household will become a compound household. Births in a family may signal that a household is in the growth phase of the domestic life-cycle, and families may wish to accommodate their growing numbers with additional dwelling space while maintaining the economic advantages of a larger labor pool. This finding is consistent with the negative effect of births on the dissolution of compound households. In the interaction model, the main effects of C/P ratio and births remain relatively unchanged, while the interaction of land value with household size and the main effect of land value are also significant. Households with

132 access to more land are more likely to become compound households, although the overall effect is small.

In the main effects model of the formation of extended households, deaths in the family significantly decrease the risk of household extension. This finding is unsurprising, as the deaths of household members increase the likelihood that the potential individuals who could contribute to the categorization of that household as extended, such as grandparents or grandchildren, are no longer present. In the interaction model, the effect of deaths remains the same, while the interaction term is small, but significant, and the effect of land value is small, but approaching significance (p = 0.087). Thus, when the interaction of household size and land value is considered, the amount of land available to a household slightly increases the chances that the household becomes extended.

Discussion and Future Directions

Descriptive statistics and event history models demonstrate that compound and extended households are different, both in several variables of interest and in their risks of formation and dissolution. The differential effects of births and deaths on the risks of formation and dissolution are particularly striking. The effects of deaths on extended households may be related to the fact that the categorization of a household as extended may depend on the presence of only one person. However, in compound households, the death of a single individual is less likely to affect the classification of the rest of the household as compound. The role of births, however, may indicate that a different process is at work. Compound households may be a way for growing families to adjust to increased numbers while ensuring greater privacy and living space but maintaining economic dependence and a larger labor pool. Suppose that these families are analogous to stem or joint households if it were not for the existence of separate dwellings, such

133 as in the case of South Hammer. If this is the case, compound households may be forming and splitting in a response to younger generations beginning and ending their childbearing years.

When a couple present in a household begins to have children, the household‘s need or desire for additional living space increases with the addition of new members and thereby increases the risk of transition to a compound living arrangement. Years later, when those children begin to leave home, there is a lesser need for additional dwelling space, which might contribute to the consolidation of a compound household into a non-compound extended or simple household.

However, to test this hypothesis, it is important to know the kinship relationships among the component members of compound households. While the current state of data linkage does not allow for this, such relationships could be studied in the future.

Surprisingly, land values had small effects, often significant only in interaction models.

While, theoretically, access to land is an essential determinant of smallholder economics, there may not be enough variation in holdings in Orkney to discern the effects. Alternately, the use of valuation rolls as a proxy for land size may not be sensitive to other important considerations.

For example, agricultural improvements, such as manuring and drainage, take labor and capital inputs and increase production, but do not affect the overall acreage available to a household.

Finally, C/P ratios have strong, important effects on the risk of household dissolution and formation. However, the interpretation of these effects remains difficult. Perhaps the development of other measures of household composition could better elucidate the nature of household composition rather than roughly estimate the ratio of total consumers to total producers. The measurement of household composition remains an area for further development.

134 Tables and Figures

Table 3-1. Number and proportion of household types by census year. These sample sizes and proportions are different from those presented in earlier chapters because the sample selection criteria have changed. The island of Sanday, Manses, and large farms are excluded from this sample.

Number Proportion Number Proportion Number Proportion N Year Compound Compound Extended Extended Simple Simple (Overall) 1851 162 0.39 75 0.18 174 0.42 411 1861 183 0.41 85 0.19 173 0.39 441 1871 161 0.35 102 0.22 194 0.42 457 1881 174 0.37 114 0.24 184 0.39 473 1891 135 0.30 102 0.23 211 0.47 448 1901 111 0.25 111 0.25 221 0.50 443

Table 3-2. Inhabitants of the croft of South Hammer, Westray, 1901 (from 1901 UK national census).

Inhabitants of South Hammer, 1901 Dwelling Relationship Marital Number Name to head Age Status Occupation 1 Mary Paterson Wife 58 m 1843 1 Robert Paterson Head 66 m 1835 Farmer 1 Robert Paterson Son 24 s 1877 Assisting on Farm

2 Isabella Paterson Wife 56 m 1845 2 Robert Paterson Son 29 s 1872 Fisherman 2 William Paterson Head 52 m 1849 Fisherman

3 Charles Rendall Son 13 s 1888 Scholar* 3 Janet Rendall Head 51 w 1850 Housekeeper 3 Jessie Rendall Daughter 11 s 1890 Scholar* 3 John Rendall Son 17 s 1884 Ploughman on Farm** 3 William Rendall Son 21 s 1880 Ploughman on Farm**

* Student at the local grammar school ** ploughman at Tirlot, the large estate that owned South Hammer, a part-time occupation

135

Figure 3-1. Genealogical relationships of the inhabitants of South Hammer, 1901. The numbers and dashed circles represent the three dwelling units at South Hammer. Numbers inside individual symbols are ages.

Composition of South Hammer Complex,1901

house

58 66 56 52 51

fisherman, housekeeper farmer blacksmith

24 29 21 17 13 11

ploughman ploughman farm assistant fisherman scholar scholar at Tirlot at Tirlot 1 2 3

136 Figure 3-2. Plan of the surviving structures at South Hammer, 2003. Courtesy of Dr. Timothy Murtha

3

2 1

137 Figure 3-3. OSGB Map of South Hammer, showing farm and dwelling structures, 1901.

OSGB 1:2500 1901

138 Figure 3-4. Hazard of household dissolution by census year.

0.7

0.6

0.5

0.4

0.3 Extended households Compound Households

Hazard Hazard of Dissolution 0.2

0.1

0 1851 1861 1871 1881 1891

Census Interval

Figure 3-5. Hazard of household formation by census year.

0.3

0.25

0.2

0.15 Extended households

0.1 Compound Households Hazard Hazard of Dissolution

0.05

0 1851 1861 1871 1881 1891

Census Interval

139 Table 3-3. Weighting system for consumer-producer ratios, after Hammel (2005).

Production Consumption Male Female Male Female Age Units Age Units Age Units Age Units 0-5 0 0-5 0 0-2 0.1 0-2 0.1 5-7 0.1 5-6 0.2 2-5 0.3 2-5 0.3 7-9 0.2 6-10 0.5 5-9 0.5 5-6 0.5 9-12 0.5 10-15 0.7 9-12 0.7 6-10 0.7 12-15 0.9 15-20 0.7 12-15 0.8 10-12 0.8 15-50 1 20-60 0.8 15-50 1 12-60 0.8 50+ 0.8 60+ 0.7 50+ 0.8 60+ 0.7

Table 3-4. Composition (percentage) of simple family households by census year.

HH Composition 1851 1861 1871 1881 1891 1901 No family/Solitaries 7.47 6.36 13.40 11.96 9.48 11.31 Married Head 81.61 83.24 77.32 76.09 78.67 73.76 Widower/Widow Head 10.92 10.40 9.28 11.96 11.85 14.93 Total 100.00 100.00 100.00 100.00 100.00 100.00

Table 3-5. Extended family household by type of extension (percentage) and census year.

Type of Extension 1851 1861 1871 1881 1891 1901 Up 26.67 32.94 24.51 35.09 29.41 31.53 Down 25.33 30.59 27.45 22.81 19.61 22.52 Lateral 14.67 12.94 14.71 16.67 13.73 8.11 Combination 10.67 7.06 14.71 6.14 13.73 12.61 Multiple family 22.67 16.47 18.63 19.30 23.53 25.23 Total 100.00 100.00 100.00 100.00 100.00 100.00

Table 3-6. Composition (percentage) of compound households by census year. headings describe component units of compound households. HH Composition 1851 1861 1871 1881 1891 1901 At least 1 extended unit 32.10 28.96 34.78 36.21 34.07 37.84 Only simple units 62.35 65.57 57.76 56.90 59.26 56.76 At least 1 multiple unit 5.56 5.46 7.45 6.90 6.67 5.41 Total 100.00 100.00 100.00 100.00 100.00 100.00

140 Table 3-7. Average household size by type and census year.

Year Compound Extended Simple 1851 10.48 6.37 5.30 1861 10.08 6.33 5.10 1871 10.61 6.25 4.72 1881 9.39 6.32 4.85 1891 9.55 6.10 4.58 1901 8.99 5.86 4.43

Table 3-8. Average number of vital events per census interval by household type.

Mean Number of Deaths Interval Compound Extended Simple 1851-1861 0.59 0.49 0.44 1861-1871 1.30 1.00 0.49 1871-1881 1.09 0.70 0.80 1881-1891 1.05 0.82 0.52 1891-1901 1.21 1.04 0.69 Mean Number of Births Interval Compound Extended Simple 1851-1861 2.12 1.52 1.16 1861-1871 2.54 1.34 1.32 1871-1881 2.28 1.66 0.86 1881-1891 1.95 1.24 0.96 1891-1901 1.71 1.34 1.09

Table 3-9. Mean land valuation by household type.

Year Compound Extended Simple 1851 10.55 6.58 6.97 1861 10.61 8.20 6.54 1871 12.11 9.13 7.61 1881 14.99 11.11 8.88 1891 13.49 8.62 8.59 1901 12.79 7.31 8.28

Table 3-10. Average and standard deviation of consumer-producer ratios by household type.

Mean Std. Dev. Mean Std. Dev. Mean Std. Dev. Year Compound Compound Extended Extended Simple Simple 1851 1.106 0.083 1.095 0.077 1.117 0.118 1861 1.112 0.087 1.109 0.082 1.108 0.132 1871 1.107 0.080 1.104 0.096 1.090 0.125 1881 1.097 0.082 1.099 0.095 1.088 0.122 1891 1.096 0.092 1.099 0.093 1.080 0.120 1901 1.087 0.095 1.079 0.087 1.069 0.112

141 Table 3-11. Number of households by ownership status and type. Collapsed over census years, 1851-1901.

Ownership Number Number Number Number Mean HH Mean Status Households Simple Extended Compound Size Valuation Landowner 60 32 8 19 6.78 11.41 Laird's Tenant 1,954 797 459 698 7.07 9.47 Other Tenant 48 17 7 24 7.52 17.69

Table 3-12. Estimates of covariate effects on household dissolution, stratified by household type, discrete-time logit hazard model, Northern Orkney, 1851-1901. Main effects and interaction models reported. Standard errors adjusted for multiple observations by clustering over household identifiers. The number of clusters in Household ID are different from the sample size reported in Table 3-1 and the number in Table 3-13 because the specific households at risk of dissolution and formation change over time.

Main Effects Interaction Coef. p-value Coef. p-value Extended Households Household Size -0.1166 0.050 -0.0978 0.214 Land Value 0.0323 0.040 0.0473 0.347 C-P Ratio 3.3389 0.001 3.4347 0.001 Landowner (1=yes, 0=no) 1.0355 0.134 1.0691 0.132 Number of Births -0.0958 0.184 -0.0966 0.181 Number of Deaths 0.3712 0.002 0.3690 0.002 Land Value X Household Size -0.0019 0.737 Time -0.0980 0.264 -0.0973 0.268 Compound Households Household Size -0.1008 <.001 -0.0433 0.234 Land Value -0.0065 0.468 0.0485 0.062 C-P Ratio 3.3324 <.001 3.0482 0.001 Landowner (1=yes, 0=no) 0.9054 0.136 1.0345 0.112 Number of Births -0.1741 <.001 -0.1807 <.001 Number of Deaths -0.0300 0.696 -0.0294 0.703 Land Value X Household Size -0.0051 0.024 Time 0.0560 0.376 0.0557 0.377 Constant -2.9351 0.001 -3.1768 0.001 Log-likelihood -633.2765 -630.0251 BIC 1370.154 1377.465 Observations 999 999 Clusters in Household ID 361 361 Pseudo R-Square 0.074 0.078

142 Table 3-13. Estimates of covariate effects on household formation, stratified by household type, discrete-time logit hazard model, Northern Orkney, 1851-1901. Main effects and interaction models reported. Standard errors adjusted for multiple observations by clustering over household identifiers. The number of clusters in Household ID are different from the sample size reported in Table 3-1 and the number in Table 3-12 because the specific households at risk of dissolution and formation change over time.

Main Effects Interaction Coef. p-value Coef. p-value Extended Households Household Size -0.0226 0.274 0.0128 0.595 Land Value -0.0057 0.488 0.0263 0.087 C-P Ratio -1.6025 0.004 -1.6675 0.003 Landowner (1=yes, 0=no) -0.3523 0.432 -0.3220 0.474 Number of Births 0.0162 0.653 0.0131 0.719 Number of Deaths -0.1598 0.031 -0.1636 0.029 Land Value X Household Size -0.0034 0.012 Time 0.0132 0.804 0.0139 0.794 Compound Households Household Size 0.0056 0.908 0.0975 0.120 Land Value 0.0150 0.200 0.0732 0.001 C-P Ratio -1.5378 0.020 -1.8504 0.008 Landowner (1=yes, 0=no) -0.1004 0.835 -0.0449 0.928 Number of Births 0.2164 <.001 0.2169 <.001 Number of Deaths 0.0505 0.589 0.0351 0.709 Land Value X Household Size -0.0083 0.013 Time -0.2918 <.001 -0.2940 <.001 Constant 0.5558 0.346 0.3482 0.567 Log-likelihood -1047.655 -1041.494 BIC 2211.203 2214.333 Observations 2267 2267 Clusters in Household ID 426 426 Pseudo R-Square 0.032 0.038

143 References

Alderson, A.S. and S.K. Sanderson. 1991. "Historic European household structures and the capitalist world-economy." Journal of Family History 16(4):419-432.

Alter, G. 1988. Family and the Female Life Course: The Women of Verviers, Belgium, 1849- 1880. Madison: University of Wisconsin Press.

Anderson, M. 1971. Family Structure in Nineteenth Century Lancashire. Cambridge: Cambridge University Press.

—. 1995. Approaches to the History of the Western Family, 1500-1914. Cambridge: Cambridge University Press.

Berkner, L.K. 1972. "The stem family and the developmental cycle of the peasant household: An eighteenth-century Austrian example." The American Historical Review 77(2):398.

—. 1975. "The use and misuse of census data for the historical analysis of family structure." Journal of Interdisciplinary History 5(4):721-738.

Boulton, J. 2000. "'It is extreme necessity that makes me do this': Some "survival strategies" of pauper households in London's West End during the early eighteenth century." Pp. 47-70 in Household Strategies for Survival, 1600-2000: Fission, Faction, and Cooperation, edited by L. Fontaine and J. Schlumbohm. Cambridge: Cambridge University Press.

Burch, T.K. 1970. "Some demographic determinants of average household size: An analytic approach." Demography 7(1):61-69.

Carter, A.T. 1984. "Household histories." Pp. 44-83 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

Chayanov, A.V. 1986. The Theory of Peasant Economy. Madison: University of Wisconsin Press.

Coale, A.J. 1965. "Appendix: Estimates of average size of household." in Aspects of the Analysis of Family Structure, edited by M. Levy and A.J. Coale. Princeton: Princeton University Press.

Cory, K.B. 2004. Tracing Your Scottish Ancestry. Edinburgh: Birlinn.

Czap, P. 1982. "The perennial multiple family household, Mishino, Russia 1782-1858." Journal of Family History 7(1):5-26.

—. 1983. "'A large family: the peasant's greatest wealth': serf households in Mishino, Russia, 1814-1858." Pp. 105-152 in Family Forms in Historic Europe, edited by R. Wall, J. Robin, and P. Laslett. Cambridge: Cambridge University Press.

144 Das Gupta, M. 1997a. "Kinship systems and demographic regimes." Pp. 36-52 in Anthropological Demography: Toward a New Synthesis, edited by D.I. Kertzer and T. Fricke. Chicago: University of Chicago Press.

—. 1997b. "Socio-economic status and clustering of child deaths in rural Punjab." Population Studies 51(2):191-202.

—. 1999. "Lifeboat versus corporate ethic: Social and demographic implications of stem and joint families." Social Science and Medicine 49(2):173-184.

Elder, G.H. 1978. "Family history and the life course." Pp. 17-64 in Transitions: The Family and the Life Course in Historical Perspective, edited by T.K. Hareven. New York: Academic Press.

Erdozáin-Azpilicueta, P. and F. Mikelarena-Peña. 1998. "Labor power, social and economic differentials, and adaptive strategies of peasant households in stem-family regions of Spain." The History of the Family 3(2):155-172.

Fjellman, S.M. 1977. "The Akamba domestic cycle as Markovian process." American Ethnologist 4(4):699-713.

Flandrin, J.L. 1976. Families in Former Times: Kinship, Household and Sexuality. Cambridge: Cambridge University Press.

Fontaine, L. and J. Schlumbohm. 2000. "Household strategies for survival: An introduction." Pp. 1-18 in Household Strategies for Survival, 1600-2000: Fission, Faction, and Cooperation, edited by L. Fontaine and J. Schlumbohm. Cambridge: Cambridge University Press.

Fortes, M. 1958. "Introduction." Pp. 1-14 in The Developmental Cycle in Domestic Groups, edited by J. Goody. Cambridge: Cambridge University Press.

Gaunt, D. 1978. "Household typology: Problems--methods--results." Pp. 69-83 in Change and Change: Social and Economic Studies in Historical Demography in the Baltic Area, edited by S. Akerman, H.C. Johansen, and D. Gaunt. Odense: Odense University Press.

Goody, J. 1996. "Comparing family systems in Europe and Asia: Are there different sets of rules?" Population and Development Review 22(1):1-20.

Goubert, P. and G. Denault. 1977. "Family and province: A contribution to the knowledge of family structures in early modern France." Journal of Family History 2(3):179-195.

Hagen, E.H., H.C. Barrett, and M.E. Price. 2006. "Do human parents face a quantity-quality tradeoff?: Evidence from a Shuar community." American Journal of Physical Anthropology 130(3):405-418.

Hajnal, J. 1965. "European marriage patterns in perspective." Pp. 101-143 in Population in History, edited by D.V. Glass and D.E.C. Eversley. London: Arnold.

145 —. 1982. "Two kinds of preindustrial household formation systems." Population and Development Review 8(3):449-494.

Hammel, E.A. 1972. "The zadruga as process." Pp. 335-373 in Household and Family in Past Time, edited by P. Laslett and R. Wall. Cambridge: Cambridge University Press.

—. 2005. "Chayanov revisited: A model for the economics of complex kin units." Proceedings of the National Academy of Sciences 102(19):7043-7046.

Hammel, E.A. and P. Laslett. 1974. "Comparing household structure over time and between cultures." Comparative Studies in Society and History 16(1):73-109.

Janssens, A. 2002. Family and Social Change: The Household as a Process in an Industrializing Community. Cambridge: Cambridge University Press.

Jennings, J.A., C.S. Sparks, J.W. Wood, P.L. Johnson, T.M. Murtha, and S.A. Matthews. 2009. "Formation and dissolution of extended-family households in Orkney, Scotland, 1851- 1901." Presented at the meeting of the International Union for the Scientific Study of Population, Marrakech, Morocco.

Jennings, J.A. and J.W. Wood. 2009. "Measuring Consumer-to-Producer Ratios: Do Weighting Systems Matter?" Presented at the annual meeting of the Social Science History Association, Long Beach, CA.

Katz, M.B. 1975. The People of Hamilton, Canada West: Family and Class in a Mid-nineteenth- Century City. Cambridge: Harvard University Press

Kertzer, D.I. 1984. Family Life in Central Italy, 1880-1910: Sharecropping, Wage Labor, and Coresidence. New Brunswick: Rutgers University Press.

—. 1989. "The joint family household revisited: Demographic constraints and household complexity in the European past." Journal of Family History 14(1):1-15.

Kunstadter, P. 1984. "Cultural ideals, socioeconomic change, and household composition: Karen, Lua', Hmong, and Thai in Northwestern Thailand." Pp. 299-329 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

Laslett, P. 1977. "Introduction." Pp. 23-89 in Household and Family in Past Time, edited by P. Laslett and R. Wall. Cambridge: Cambridge Univ. Press.

—. 1984a. "The family as a knot of individual interests." Pp. 353-382 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

—. 1984b. The World We Have Lost New York: Scribner.

—. 1988. "Family, kinship and collectivity as systems of support in pre-industrial Europe: A consideration of the ‗nuclear-hardship‘ hypothesis." Continuity and Change 3(2):153- 175.

146 Laslett, P., J. Robin, and R. Wall. 1983. Family Forms in Historic Europe. Cambridge: Cambridge University Press.

Laslett, P. and R. Wall. 1974. Household and Family in Past Time. Cambridge: Cambridge University Press.

—. 1977. Household and Family in Past Time. Cambridge: Cambridge University Press.

Le Play, F. 1982. On Family, Work, and Social Exchange. Chicago: University of Chicago Press.

Mitterauer, M. and R. Sieder. 1979. "The developmental process of domestic groups: Problems of reconstruction and possibilities of interpretation." Journal of Family History 4(3):257- 284.

Murtha, T.M., P.L. Johnson, J.W. Wood, S.A. Matthews, C.S. Sparks, and J.A. Jennings. 2008. "Historical demography, oral history, settlement archaeology, and landscape ecology: The North Orkney Population History Project." Presented at the annual meeting of the European Population Conference, Barcelona, Spain.

Netting, R.M. 1993. Smallholders, Householders: Farm Families and the Ecology of Intensive, Sustainable Agriculture. Stanford: Stanford University Press.

Otterbein, K.F. and C.S. Otterbein. 1977. "A stochastic process analysis of the developmental cycle of the Andros household." Ethnology 16(4):415-425.

Palsson, H. and P. Edwards. 1981. Orkneyinga Saga: The History of the Earls of Orkney. London: Penguin Books.

Plakans, A. and C. Wetherell. 2005. "The Hajnal line and Eastern Europe." Pp. 105–128 in Marriage and the Family in Eurasia: Perspectives on the Hajnal Hypothesis, edited by T. Engelen and A.P. Wolf. Amsterdam: Aksant Academic.

Reyna, S.P. 1976. "The extending strategy: Regulation of the household dependency ratio." Journal of Anthropological Research 32(2):182-198.

Rudolph, R.L. 1992. "The European family and economy: central themes and issues." Journal of Family History 17(2):119-138.

Ruggles, S. 1987. Prolonged Connections: The Rise of the Extended Family in 19th Century England and America. Madison: University of Wisconsin Press.

Segalen, M. 1977. "The family cycle and household structure: five generations in a French village." Journal of Family History 2(3):223-236.

—. 1983. Love and Power in the Peasant Family. Rural France in the Nineteenth Century. Oxford: Blackwell.

—. 1984. "Nuclear is not independent: Organization of the household in the Pays Bigouden Sud in the nineteenth and twentieth centuries." Pp. 163-186 in Households: Comparative and

147 Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

Sieder, R. and M. Mitterauer. 1983. "The reconstruction of the family life course: theoretical problems and empirical results." Pp. 309-346 in Family Forms in Historic Europe, edited by R. Wall, J. Robin, and P. Laslett. Cambridge: Cambridge University Press.

Skinner, G.W. 1997. "Family systems and demographic processes." Pp. 53-95 in Anthropological Demography: Toward a New Synthesis, edited by D.I. Kertzer and T. Fricke. Chicago: University of Chicago.

Spagnoli, P.G. 1983. "Industrialization, proletarianization, and marriage: a reconsideration." Journal of Family History 8(3):230-247.

Sparks, C.S. 2007. Households, Land, and Labor: Population Dynamics in the Northern Orkney Islands, Scotland, 1851 to 2003. Ph.D. Dissertation, Anthropology and Population Research Institute, Pennsylvania State University.

—. 2009. "An application of the variable-r method to subpopulation growth in a 19th century agricultural population." Demographic Research 21:23-64.

Szoltysek, M. 2007. "Central European household and family systems, and the ‗Hajnal– Mitterauer‘ line: The parish of Bujakow (18th–19th centuries)." The History of the Family 12(1):19-42.

The National Archives of Scotland. 2003. Tracing Your Scottish Ancestors, The Official Guide: A Guide to Ancestry Research in the National Archives of Scotland. Edinburgh: Mercat Press.

Thomson, W.P.L. 2008a. The New History of Orkney. Edinburgh: Birlinn Ltd.

—. 2008b. Orkney Land and People. Kirkwall: The Orcadian Limited (Kirkwall Press).

Van de Walle, E. 1976. "Household dynamics in a Belgian village, 1847-1866." Journal of Family History 1(1):80-94.

Viazzo, P.P. 2005. "South of the Hajnal line: Italy and Southern Europe." Pp. 129-164 in Marriage and the Family in Eurasia: Perspectives on the Hajnal Hypothesis, edited by T. Engelen and A.P. Wolf. Amsterdam: Aksant.

Viazzo, P.P., M. Aime, and S. Allovio. 2005. "Crossing the boundary: Peasants, shepherds, and servants in a western Alpine community." The History of the Family 10(4):387-405.

Wakefield, D. 1998. Fenjia: Household Division and Inheritance in Qing and Republican China. Honolulu: University of Hawai'i Press.

Wall, R. 1983. "Introduction." Pp. 1-64 in Family Forms in Historic Europe, edited by R. Wall, J. Robin, and P. Laslett. Cambridge: Cambridge University Press.

148 Wilk, R. 1984. "Households in process: Agricultural change and domestic transformation among the Kekchi Maya of Belize." Pp. 217-244 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

Wilk, R.R. 1991. Household Ecology: Economic Change and Domestic Life Among the Kekchi Maya in Belize. Tucson: University of Arizona Press

Wilk, R.R. and R.M.C. Netting. 1984. "Households: Changing Forms and Functions." Pp. 1-28 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R.R. Wilk, and E.J. Arnould. Berkeley: University of California Press.

Wolf, A.P. 1984. "Family life and the life cycle in rural China." Pp. 279-298 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

Chapter 4

Household Predictors of the Presence of Life-cycle Servants in Northern Orkney, 1851-1901

On the 16th October in the year 58, I went to the ploo, nae doots to haud straucht, Mysel‘ in good humour, my horses the same, I ploo‘d till eleiven, and then I cam‘ hame, Gaed Nell and Nansy some corn to eat, An‘ syne took a besom and swypet my feet. Then to the barn I quickly withdrew, To bundle some straw, but, oh, what a stew! When denner was over to the stable we went, To clean up oor horses it was oor intent. Oor order fae Dawson for plooin, again, Nae sinsheen, but cloody and some draps o‘ rain. Noo my day‘s wark is finisht, and I‘ll hae a smoke, An‘ I‘m boun for my bed, for it‘s past nine o‘clock. (Buchan 1984:238)18

Introduction

The significance of the presence of servants in Northwest European households was first recognized by historical demographers in their description of what has come to be known as the

―European Marriage Pattern‖ (Hajnal 1965; Laslett 1977). In his influential paper on rules of household formation, Hajnal (1982) indentified as a characteristic of this region in the 17th-19th centuries the frequent circulation of young people as household servants. He described the general nature of the institution of service as follows:

18 Poem written in 1858 on the wall of a chaumer in Aberdeenshire. A chaumer is a structure attached to the main farming buildings and is often a loft. It customarily houses agricultural servants, but unlike a bothy, a separate structure to house unmarried male farm laborers, it is not an outbuilding. Fae=from, ploo=plow, besom=broom, corn=grain (probably oats). Nell and Nansy are the horses and Dawson is presumably the landowner or foreman/factor.

150 (1) Servants were numerous, apparently always constituting at least 6 per cent, and usually over 10 per cent, of the total population. (2) Almost all servants were unmarried and most of them were young (usually between 10 and 30 years of age. (3) A substantial proportion of young people of both sexes were servants at some stage in their lives. (4) Most servants were not primarily engaged in domestic tasks, but were part of the work force of their master‘s farm or craft enterprise. (5) Servants lived as members of their master‘s household. (6) Most servants were members of their master‘s household by contract for a limited period. (7) There was no assumption that a servant, as a result of being in service, would necessarily be socially inferior to his or her master. The great majority of servants eventually married and ceased being servants. Their social class before service (i.e., usually the class of their parents) and their social class after service could be the same as their master‘s (and in some Northwest European populations at some periods this was not infrequently the case). (Hajnal 1982:473)

Laslett termed these young laborers ―life-cycle servants‖ in acknowledgement of the status of these individuals as temporary servants, rather than as members of a servant class

(1977:34). Indeed, it is essential to think of these servants in life-cycle terms, as their status was not permanent, and although service provided useful training in agricultural tasks, it was not a career (Goldberg 1992; Whittle 2000). Life-cycle service is commonly linked to the system of household formation that features late age at marriage, high rates of celibacy, and the predominance of nuclear households, but research outside of Northwest Europe, particularly in

Italy, has provided examples of both nuclear household systems that do not feature life-cycle servants, and multiple (or complex) household systems that do (Arru 1990; Kertzer 1984; Kertzer and Barbagli 2002; Kertzer and Hogan 1990; Landsteiner 1999; Molin 1990). Life-cycle service has also been documented in Japan, far east of the ―Hajnal line‖ that is said to define the spatial extent of the Northwest European marriage pattern (Nagata 2005). For the purposes of this study, the description and discussion of the institution of life-cycle service will be restricted to how it was practiced in Northwest Europe, particularly in the British Isles.

This study examines household predictors of the number of life-cycle servants present within households. It also compares predictors of the number of male and female servants.

151 Poisson regression methods are used to assist in the identification of factors that influence the number and sex of servants present in the household, after controlling for various household characteristics, including those related to the household head.

Servants are often viewed as a supplement to household labor employed to ensure sufficient food and craft production and full use of the household‘s resources, such as land and livestock. Therefore, all other things being equal, it is hypothesized that the amount and quality of land to which a household has access, the size and composition of the household labor force, and the overall consumption requirements of the household will be important predictors of the number of servants a household hires. In addition, male and female labor may be deployed differently, depending on the specific tasks a household must accomplish. Thus, the sex of the servant hired might be predicted by household factors, including access to land, household age and sex composition, and household size, while controlling for other factors.

Data from Orkney, off the northern coast of mainland Scotland, provide an opportunity to contribute to the study of life-cycle service, an important part of life in Europe in the preindustrial past. Service occupied many young people, moved people over the landscape, and was often an important component of the household economy of traditional farmers. Service in Orkney persisted throughout the study period (1851-1901) and available historical sources provide information about individuals, households, and landholdings, making this dataset useful for the examination of how household composition and land resources affected the hiring of servants.

152 Background

Setting & Data Sources

The Orkney Islands lie off the north coast of Scotland where the North Sea and

Atlantic Ocean meet. Six islands, Westray, Sanday, Papa Westray, Eday, Faray, and North

Ronaldsay, are included in the study area. These islands are the focus of an ongoing multidisciplinary study of population, family history, settlement, and land use known as the North

Orkney Population History Project (Murtha et al. 2008; Sparks 2007). The islands are still rural in character and during the period of interest (1851-1901) featured mixed agricultural production based on grains, mainly black oats and bere (a landrace of barley), root crops, such as potatoes and turnips, and livestock, including cattle, sheep, chickens, and pigs. In the study period, these activities supported household subsistence and paid rents. Individual farmsteads, or groupings of household and agricultural structures with associated gardens and fields, were the primary locus of agricultural production, and were dispersed over the islands, with a few small villages that were more densely settled but still rural in character. The farmsteads all have names, and these names appear in historical records and persist for many generations, even though the inhabitants change over time (Palsson and Edwards 1981; Thomson 2008a, 2008b). Estate maps from the

1830‘s and 1840‘s feature farmsteads with the same names and locations as in later historical and modern maps. Orkney historians speculate that most of the farm names became fixed by the

1840‘s to accommodate record keeping, such as census enumerations and valuation rolls, and mail delivery (Thomson 2008a). The population of the islands reached its height in the mid- to late-1800s, after which out-migration, both to mainland Scotland and overseas, and declining fertility initiated the depopulation that continues to this day.

153 Farmstead names are listed in decennial census returns, along with the inhabitants of the farmstead and their relationship to the household head. This allows for the identification of servants, and the tracking of household size and composition over time. Valuation rolls, which are records of property values on which taxes were assessed, are linked to the census returns using farmstead names. These records are a good proxy measure of landholding size and quality.

This assessment is based on a sample of landholdings for which acres in pasture and arable could be obtained from surviving cadastral (estate) maps. In this sample, the amount of valuation, is almost entirely a reflection of the sizes of holdings (R2=0.98).19 The linked census and valuation records provide information about the household and household head. Many measures, such as valuation, household size, and age of the head, are taken directly from the historical records. Two theoretically important measures, household consumer-to-producer (C/P) ratio and household type, are the result of more elaborate calculation and classification and will be taken up in turn.

C/P Ratios

Traditional farming systems are characterized by human labor inputs, limited use or lack of fossil fuels (for tractors, for instance), limited use of wage laborers, and household organization of production (Netting 1993; Redfield 1989; Wolf 1966). Chayanov (1986) noted that the absence of household labor distinguished smallholders from large-scale capitalist farming firms. In his theory of the peasant economy, the balance of workers relative to consumers was essential to understanding how families meet their needs while minimizing the drudgery of additional labor. Chayanov described how the relationship of household labor supply to consumption needs changes as the family develops. In his simple model, the consumer-producer

(C/P) ratio is followed from the formation of a new household by a couple through their

19 See Appendix A for details.

154 childbearing years (Table 4-1). As children are born, the C/P ratio becomes sharply unfavorable

(Figure 4-1), placing economic pressure on the household that may be mediated by increased intensity of production, decreased consumption, or the acquisition of additional labor, including the hiring of life-cycle servants. In his consideration of the changing nature of household workers and consumers, Chayanov applied a set of age- and sex-based weights to each household member, meant to represent their relative contributions to household production and their nutritional needs. Other studies, expanding on Chayanov‘s original scheme, have adopted their own sets of age- and sex-specific weights (Chibnik 1984; Hammel 2005; Hunt 1979; Lewis

1981). In this study, an updated set of weights proposed by Hammel in an elaboration of

Chayanov‘s model is used (Table 4-2).20 While the choice of weights may affect the result of statistical analyses, in a comparison of various measures of age- and sex-related weighting systems, the specific choice of weighting system have little effect on the coefficient of the C/P ratio variable when used as a predictor in simple regression models applied to data from this study area (Jennings and Wood 2009).

Regardless of the weighting system chosen, several studies find important effects of C/P ratios. Infant mortality is sensitive to increasing C/P ratios, so that families with unfavorable economic conditions are more likely to experience the death of an infant (Campbell and Lee

1996; Sparks 2007). These findings are corroborated by evidence that children‘s anthropometric measurements, such as height, weight, and body fat, are lower for their age in families with unfavorable C/P ratios (Hagen et al. 2006). Others have found that household land use allocation changes over the course of the family life cycle, a concept analogous to Chayanovian cycles in

C/P ratios (Perz 2003). C/P ratios are also predictive of the formation and dissolution of extended-family living arrangements (Jennings et al. 2009a, 2009b). Given the importance of

20 Chayanov‘s original weights were substituted in the regression models described below, and the results did not substantively change. See Appendix B for more details on different C/P ratio measurements and their effects on model coefficients.

155 C/P ratios to various dimensions of the household economy and the demographic fortunes of households, it is hypothesized here that C/P ratios will also be an important predictor of the number and sex of household servants, who, as young adult household members, can have a considerable influence on the balance of household laborers relative to household consumption needs.

Household Types

Hajnal (1982) links life-cycle service to a household formation system in which nuclear

(simple) households were the dominant form. Simple households consist of a married couple, or a surviving member of that couple, and their children.21 Extended households, in contrast, feature one or more individuals additional to the simple household. In this study, extended households refer to both extended households and multiple households, which include two or more married couples, as outlined in the classification system of Hammel and Laslett (1974). A third household type, not included in this classification system, is found in Orkney (Jennings et al.

2009a, 2009b). Using archaeological evidence, in tandem with written records, the North Orkney

Population History Project has found that many units listed separately in the census were actually single economic entities. These ―compound‖ households, often linked by brothers, are adjacent, or even structurally joined, and share a common set of farming structures such as barns, byres, grain kilns, and stables. Given what is known about Orcadian economic and social systems, the component units of compound households were probably not independent. Rather, they worked their landholdings cooperatively and shared the products of their labor, even if employed in outside wage labor, which was often part-time or seasonal. With respect to life-cycle servants, it

21 These definitions refer to related individuals within the household. The presence of servants, as well as lodgers, boarders, and visitors, is not inconsistent with simple households as defined here. Thus, a household that contains a married couple, their children, a servant, and a lodger is classified as a simple household.

156 is hypothesized that compound, and possibly extended, households were less likely to hire servants, as their larger size and more complex kin organization might provide them with a larger and more diverse labor force, thereby decreasing the need for servant labor.

With these data sources and issues in mind, this paper examines the practice of life-cycle service in north Orkney in detail, paying particular attention to the characteristics of households that send members out as servants (source households) and the households that hire servants (sink households). The sexual division of agricultural labor and the different functions of male and female servants will also be considered.

Defining Servants

The precise definition of agricultural servants is complicated by the diverse, and often misleading, set of terms used to describe them. Even in her authoritative work on the subject of servants in husbandry, Kussmaul (1981b) describes servants as inherently ambiguous. Early descriptions of life-cycle servants shed some light on their defining characteristics. In his 1587 treatise on England, William Harrison, a contributor to Holinshed‘s Chronicles, an important source for Elizabethan social history, notes that English yeomen [smallholders] are ―for the most part farmers to gentlemen or at the leastwise artificers; and with grazing, frequenting of markets, and keeping of servants (not idle servants as gentlemen do, but such as get both their own and part of their master‘s living) do come to great wealth…‖ (1968:117-118). This early observer makes a distinction between ―idle‖ gentlemen‘s servants, such as butlers or grooms, and the working servants of agricultural households. In 1765, the great English jurist, William

Blackstone (1979), defines four ―species‖ of servant: domestic (life-cycle) servants, apprentices, day laborers, and factors/stewards. His description of the legal regulations regarding servants suggests their ubiquity and importance to the agricultural economy and rural life in general.

157 For this study, the definition developed by Kussmaul (1981a), who refers to life-cycle servants as servants in husbandry, is adopted. First, life-cycle servants are productive, hired to maintain the household economy, rather than maintain a lifestyle. Indeed, it has been suggested that much of the social movement of servants was lateral, instead of upward, meaning that servants worked in households similar to their own, rather than strictly for the well-to-do (Cooper

2004; Hajnal 1982; Laslett 1977). Life-cycle servants were found predominantly in rural districts and were not formal apprentices, whose parents usually paid the masters for several years of training. Generally, life-cycle servants had annual contacts, initiated during a hiring season at

Martinmas [November 11] or Whitsunday [in Scotland, May 15] (Hasbach 1908; Kitchen 1981;

Watson 1894). A contemporary account of the Martinmas or Whitsuntide hiring fair, and servants in general, is given in Appendix C.

Life-cycle servants were provided room and board as well as a wage. Given the small dwellings of most families during the period, this meant that life-cycle servants usually lived as part of their master‘s family and ate from the master‘s table (Kussmaul 1981a). Often, servants would change farms frequently, usually after a stay of only one or two years, unless both the master and servant found the situation unusually pleasant and productive (Devine 1984a;

Kussmaul 1981b; Laslett 1977). Wages varied by age and sex, with lads earning the least, and women usually earning about half as much as men (Devine 1984b; Whittle 2005). Unlike day laborers, who were hired for a day, week, or specific task, the labor of life-cycle servants was continuously available to the farmer over the period of service.

The status of the life-cycle servant was ambiguous; they might not be closely related to their master, yet they lived as part of their master‘s family and were often described in familial

158 terms (Hasbach 1908; Kussmaul 1981b).22 They were also wage earners, who were saving for marriage or sending remittances to their families. This particular function of life-cycle service is a major tenet of the ―Hajnal hypothesis‖ (Engelen and Wolf 2005). It proposes that, in Northwest

Europe, there existed a normative requirement that a couple establish an independent household upon marriage. In order to do so, young people must inherit or earn, often through life-cycle service, sufficient resources to set up a household of their own. Hajnal (1982) suggests that late age at marriage together with the requirement for newly married couples to set up their own separate households kept rural populations in balance with their productive resources, such as land. However, the extent and direction of causation, if any, among late age at marriage, life- cycle service, and neolocal post-marital residence remain unclear.

Room and board, provided to servants in addition to their wages, largely sheltered them from the variations in prices in the housing and food markets that plagued landless, or nearly landless, day laborers (Whittle 2000, 2005). Appendix D outlines the life history of Fred

Kitchen, who for parts of his life worked as an agricultural servant. His account speaks to several important characteristics and functions of life-cycle service, including frequent switching of farms, saving for marriage, and vulnerability to changes in economic and demographic fortunes.

It also features comparisons of life-cycle service to day labor and work in industry.

Life-cycle servants were found throughout preindustrial Western Europe, although their prevalence in the population varies regionally. In some areas of England, servants made up about

25% of the total population, but more commonly, only 10% of the total population consisted of servants at any one time (Kussmaul 1981a; Laslett 1977, 2000). Servants were generally rare in

Italy and Spain, comprising at times only 1% to 3% of the total population (Laslett 1977; Reher

1998; Viazzo et al. 2005). In 19th century Lancashire, 28% of rural households had servants,

22 However, it may be the case that servants often had kin ties to their masters. Future study using the Orkney data may be able to address the closeness and frequency of kin relationships in service arrangements.

159 whereas almost half of the English farming households in the Cambridge Group sample included servants (Anderson 1971; Kussmaul 1981a). In towns, the percentage of servants, many of whom were presumably not ―productive‖ servants, varied widely, but this probably reflected differences in the concentration of wealth among the urban samples (Goldberg 1992). Laslett (1977) and

Anderson (1971) both estimate that over half, and perhaps up to two-thirds of young, unmarried people could expect to go into service in England. Given the ubiquity of servants throughout

Western Europe, and particularly in the British Isles, it is unsurprising that many assert that this mass movement of young people as servants is one of the major differences between the world of preindustrial Britain and our own (Kussmaul 1981a; Laslett 1977, 2000).

A Brief History of Service in England and Scotland

The origin of life-cycle service as an institution is cloudy. Evidence from English tax records indicate that about 20% of households in 1377 had servants (Smith 1984). Some scholars speculate that the labor shortage created by the Black Death provided an incentive to manage agricultural labor in this fashion (Hasbach 1908). Others cite the earlier (13th and 14th century) decline of serfdom as the motivating factor, as it allowed newly-freed smallholders to manage their own household economies (Whittle 2000).

Whatever its origins, once it was established, the practice of life-cycle service proved to have remarkable staying power, remaining largely unchanged for hundreds of years (Kussmaul

1981a). Change finally came to the system in Southern England in the late 18th century, and gradually spread northward. A series of social and economic trends probably contributed to the end of service. The rising price of grain made payments in board and in kind more costly, and, by extension, payments in cash more economical (Hasbach 1908). Increases in population and decreases in mortality lessened the labor shortage, as farmers had more kin to call upon for labor

160 (Wall 1986).23 The age at marriage was declining, in part because real wages were rising and people had the resources to marry earlier (Cooper 2005). Agricultural improvements of the late

18th and early 19th century increased farm size and improved efficiency while decreasing the amount of labor needed to manage a farm (Devine 1984a). Near cities, the growth of industry competed for the labor of young men and women (Smout 1986). These factors contributed to the rising ratio of day laborers relative to life-cycle servants (Kussmaul 1981a). As Kussmaul puts it,

―service had been nurtured by an agrarian environment of small farms, labour shortage, and a high age at marriage; it had been enmeshed in a web of social and economic relations. When the environment changed, servants ceased being hired‖ (1981a:133).

However, the decline of service did not occur uniformly throughout the British Isles. The institution survived longer in the north of England, and especially in Scotland, where the system did not fully erode until the 1940‘s (Ewan 2004; Gray 1984). While these two regions differ in the timing of the decline of service, comparative studies indicate that the practice of life-cycle service did not differ substantially between England and Scotland (Whyte 1989; Whyte and

Whyte 1988). If this is true, then why did life-cycle service persist in Scotland? Several social and economic factors may have contributed to this pattern.

Some attribute the Scottish delay to conservative or old-fashioned practices (Hasbach

1908). However, Scottish farmers did adopt agricultural improvements, such as improved plows and the elimination of sub-tenants and cottars (Devine 1984a). We might look instead to certain

Scottish agricultural practices, necessitated by climate, which differed from those in England, particularly southern England. In England, agricultural improvements led to the creation of very large arable holdings. When the demand for wheat rose, farmers responded by planting more of it. Wheat cultivation required intense, seasonal bursts of labor (Devine 1984a). This

23 Wall, however, does not explicitly consider the role of declining fertility, which may have also affected the amount of kin labor available.

161 restructuring of the agricultural sector meant that day laborers fit farmers‘ needs better than servants (Cooper 2005). Scottish agriculture, in contrast, was characterized by mixed farming

(both arable and grazing), and different staple grains, namely oats and bere instead of wheat

(Fenton 1997; Shaw 1980). Pastoral and mixed agriculture was more conducive to live-in servants, as there was a more stable demand for labor year-round (Whittle 2000). The traditional

5-course rotation of Scottish agriculture and the importance of root crops, such as potatoes and turnips that require singling, hoeing, and so on, also demanded year-round labor (Devine 1984a).

In addition, the relatively small size of Scottish farms has been implicated in the persistence of life-cycle service. Scholars have noted that service remained longer in areas with smaller holdings and tended to disappear when those holdings consolidated (Gray 1984; Whyte and

Whyte 1988). In areas of Scotland where farms were large, farmers tended to replace servants who ate at the master‘s table with the bothy system, where the male servants lived separately from the master‘s family in a nearby out-building and cooked their own meals (Gray 1984).

Scotland and England also differed in their labor markets. Scotland‘s population did not grow as quickly as that of England, so a labor shortage in the region persisted longer (Devine

1984a). Limited transportation networks, especially in the Highlands and Islands, made payment in kind and the boarding of servants more efficient, as markets and villages could be a considerable distance away. A final contributor to the persistence of life-cycle service in

Scotland was the poor law. In Scotland, unlike England, if someone was able-bodied and unemployed, they were not eligible for poor relief (Devine 1984a). This encouraged the practice of life-cycle service, as it provided both shelter and employment in a region where the poor laws could not be depended on if day labor could not be obtained.

In light of these considerations, it is not surprising that life-cycle service remained an important feature of life in Orkney throughout the 19th Century. Servants were found in the households of both large landowners and small tenants. Relative to the mainland, Orkney had

162 small, fragmented holdings, with a large number of tenants, rather than owner-occupiers (Shaw

1980). Mixed farming was practiced throughout the islands, providing comparatively even seasonal demands for labor. The islands also featured lower wages than did other regions of

Scotland, providing yet another incentive to secure payment in kind (Levitt and Smout 1984).

Some functions of life-cycle service

Although life-cycle servants earned wages, their labor was a more like a substitute for household labor than that of day-laborers, as servants were provided room and board and lived and worked alongside household members (Whittle 2000). Indeed, it is possible to think of service as a solution to a variety of problems that faced the smallholder agrarian household economy. Life-cycle service provided a means for farming families to secure labor independently from their particular demographic fortunes. Smith noted that ―service is indeed a remarkably efficient means of temporarily redistributing labour for maximum productivity‖ (1984:38). In this sense, service can be seen as a solution to the cyclical labor shortages created by changes in family composition, as described by Chayanov (1986), who in his original statement of the peasant economy, assumed that non-kin labor was unavailable or unobtainable. This is true for both source and sink families, as servants could be hired or young family members could be sent out in order to bring the family labor force into balance with the land and other available productive resources. C/P ratios, therefore, are likely to be important predictors of the number of life-cycle servants present in a household.

Servants also provided a ready means to replace essential family members lost through death or illness, a loss that might have otherwise crippled the household economy. This insurance against loss also applied to the servants, who could be protected against orphanhood through active employment in service and the food and shelter it provided (Fauve-Chamoux and

163 Wall 2005; Laslett 1988). By hiring live-in servants, employers gained control over the reliability of their labor force. They generally fed their servants well, but worked them hard. Servants were available 24 hours a day, should a crisis or other sudden demand arise, and they were kept under close supervision (Whittle 2005). Indeed, it has been hypothesized that this close supervision could have been a mechanism to check the unruliness of young laborers and maximize their productive potential by placing them under the scrutiny of employers rather than parents (Stone

1979).

For young people, service had several attractive features. Although they were still under the authority of their masters, service allowed young people independence from their parents during the period between adolescence and marriage. As servants, they were able to accumulate some wealth and useful skills before they established families of their own. However, it has been debated whether it was in fact possible for servants to save enough to purchase a holding without contributions from an inheritance, or if they were still largely dependent on either inheritance alone or some combination of inheritance and earnings (Orr 1984; Watson 1894; Whittle 2005).

Whittle (2005) estimates that in 16th century Sussex and Norfolk, the combined savings of a male and female servant over five years could purchase and furnish a cottage with a small holding (1 acre), while about 10 years of savings were needed to purchase a farmhouse with a 10 acre holding. Although dependent on their masters, servants were largely immune to the direct effects of fluctuations in market prices and employment shortages that could ruin the prospects of a day laborer (Whittle 2000). Instead, market effects were probably passed on to them indirectly in terms of factors such as food supply or demand or more intensive labor. In fact, it may have been that for many servants, leaving the occupation entailed a decrease in their standard of living, as they had to purchase food and lodging on the open market (Whittle 2005). Finally, servants‘ mobility allowed them to establish social networks beyond their kinship and local ties (Ewan

2004).

164 Wall (1986) argues that these features made service part of the ―adaptive family economy.‖ The institution of service provided flexibility to social, familial, and economic relationships and allowed households to diversify the employment of their members. Yet, while life-cycle service seems to be an elegant solution for families facing cyclical labor shortages and surpluses and young people delaying marriage, it is important to note that the institution of service had its drawbacks. Abuses occurred on the part of both master and servant, as evidenced by legal documents and complaints of the day (Blackstone 1979; Kussmaul 1981a; Whittle 2005).

These include, but were certainly not limited to, disputes over wages and the quality of room and board, unreasonable labor demands, refusal or poor quality of work, theft, and leaving or terminating service without notice. The nature of the relationship between servants and their employers depended in large part upon the individual temperaments involved, and particular living and working conditions could range from very good to very bad. Informal and kin networks were probably important, not only for farmers finding servants, or servants finding positions, but also for inquiring into the character of the individuals involved (Goldberg 1992).

Indeed, according to Kussmaul (1981a), many servants were found to be working for extended kin, and parents and other kin often influenced hiring. In his autobiography, Fred Kitchen describes how he found work through the informal networks of fellow servants as well as farmers

(Kitchen 1981).

Given their important role in the household economy, the number of servants present in a household could be influenced by the nature of the household labor force, as summarized by the

C/P ratio, household size, the size of landholdings, and characteristics of the household head, such as employment in agriculture, age, sex, and marital status. Interactions of these variables, particular those related to the household labor force (C/P ratio and household size) and the size of landholdings, as measured by valuation, may prove important, as the literature on servants

165 stresses the adaptive nature of service and its function of reshaping, or perhaps even optimizing, the household labor force relative to its productive resources.

Women in Service

The Orkney data not only provide information about the number of servants present in households, but the sex of the servant is also listed in the census records. In this sample population, male and female servants are found in roughly equal numbers. However, it is not clear whether male and female servants participated in the same productive tasks, or if servants were hired for sex-specific tasks. One might imagine that certain household characteristics, indeed many of the same characteristics described above, might be differentially associated with the numbers of male and female servants. Put another way, households with certain characteristics may be more likely to hire female servants, while other households may favor male servants, depending on their task-specific labor requirements. Previous studies of the sexual division of agricultural labor provide insight into the kinds of tasks that may have been preferentially performed by one sex over the other.

In rural districts throughout the British Isles, life-cycle service was the most common employment for single women (Pinchbeck 1969). Yet, there has been some debate about the extent of women‘s participation in agriculture (Whittle 2005). Some studies focus on their work as dairy maids and cooks, while others emphasize the range of tasks women performed.

Contemporary sources, such as the Royal Commission on Labour, note:

At many branches of farm labour, a good girl will do more than an average man, yet she had to be content with half his wages. No doubt women‘s wages have doubled within the last 40 years, but the fact remains that often when working

166 side by side with the orra men and hinds she is doing as much as a man, and yet only getting half a man‘s wages.24

Differences between the sexes in the timing of leaving home (often for service) or in the division of labor have been observed in some studies. In a model of leaving the parental home,

Dribe (2000) finds that the effects of other variables on leaving home differ between the sexes, although the overall pattern is similar. Many agricultural tasks were considered either sex- specific or were dominated by one sex. During the harvest, women were frequently hired as day laborers, and worked alongside both male and female life-cycle servants. Harvesting using the sickle, a semi-circular blade attached to a short handle, was acceptable for both sexes, but harvesting using the scythe, a long blade attached to a long handle, was solely the domain of men

(Fenton 1976; Howatson 1984). Indeed, the development and adoption of heavier tools, such as the scythe, are said to be reflected in changes in the sexual division of labor (Snell 1985).

Examples of tasks commonly described as women‘s work included dairying, poultry keeping, vegetable growing, brewing, baking, weeding, and harvesting (Whittle 2000). In the context of Scottish agriculture, the digging of potatoes and turnips was a common female task

(Pinchbeck 1969). Despite these gendered activities, or the purported sexual division of labor, many studies note that female servants participated in a full range of farm work, both ―in and out‖ of the house (Gray 1984). This meant that women took part in heavy work as well as indoor duties. Others emphasize that women took part in every farm task, except the tending of horses

(Devine 1984b). This included the heaviest kinds of agricultural labor: plowing and harrowing

(Pinchbeck 1969). Yet certain tasks, especially pulling turnips, dairying, and byre work, were dominated by women (Devine 1984b). Turnips were thought to be particularly demanding of women‘s labor (Devine 1984b).

24 Royal Commission on Labour, 1893, p. 117, quoted from Devine 1984b, p. 119. Orra men work odd- jobs and hinds are farm laborers.

167 Was the labor of male and female life-cycle servants interchangeable, or did farmers hire a particular sex based on household composition, or the specific labor needs of the farmstead?

Some evidence from the cottage textile industry in Ireland suggests that the hiring of female servants, who specialized in spinning, and male servants, who specialized in weaving, depended on the age- and sex-composition of the household, as well as how much yarn was bought rather than produced by the household (Gray 2006). In addition, studies of English servants indicate that the sex composition and number of servants was determined by the status and occupation of the household head (Goldberg 1992).

This study seeks to test the null hypothesis that household-related variables are equally associated with the number of male and female servants. These models include variables potentially important to the demand for male or female labor. For example, the occupation of the head may be an important indicator, as farmers may require more male labor, but tailors may require more female labor. Valuations provide insight into the size and quality of landholdings, which may place additional sex-specific demands on labor. The sex of the household head may be predictive of the sex of servants present, as a female-headed household may prefer male labor to replace the labor of the absent (usually deceased) male head.

Service in Orkney

The institution of life-cycle service was established in Orkney by the 17th Century, when records show that there were 3-4 servants available to every 10 households (Flinn et al. 1977).

Evidence from the current study indicates that individuals identified as ―servants‖ in Orkney had characteristics consistent with life-cycle service as described above. Individuals listed as servants in census returns were predominantly young men and women, 86 percent of whom were between the ages of 12 and 30. In this period, Orkney was characterized by late age at marriage, with the

168 singulate mean age at marriage (SMAM) for females ranging from 29 to 32, and the SMAM for males ranging from 30 to 34 (Sparks 2007). SMAM is a measure of the average age at marriage computed using the proportion of single people by age, and can be interpreted as the expected number of years lived in the single state for those that marry before age 50 (Hajnal 1953). The mean age at marriage observed from the vital registration data was approximately four to five years lower for both sexes, which is probably attributable to the rate of celibacy or the out- migration of individuals of marriageable age. Servants were present in households of all family types (simple, extended, and compound) and in households throughout the range of landholding values, as measured by valuation rolls. In addition, the majority of servants were born within the study area, or elsewhere in Orkney, indicating that migration for service, while common and an important determinant of age-specific patterns of migration, was a highly localized process (Table

4-3). There also appears to have been a sex-specific pattern of migration, as females seem even more localized than males.

The Orkney data present an opportunity to advance the study of life-cycle service. The dataset includes information about both individuals and the households they live in. The first part of this study examines household-level predictors of the number of life-cycle servants. The multiple regression models presented below are designed to address issues related to the function of servants within the household economy. Specifically, do variables related to the household labor pool and consumption requirements, such as the consumer-producer (C/P) ratio, household size, and size and quality of landholdings, predict the number of servants present in the household after controlling for other factors? The second part of this study takes up the issue of the possible effects on employment of servants of the sexual division of labor in farming households. Here, the number of male servants is compared to the number of female servants using seemingly unrelated regression to adjust for the simultaneous estimation of two Poisson models (one for males and one for females) in an effort to determine whether household-level variables associated

169 with the household economy have different effects on the sex of number of servant hired.

Finally, this study makes a preliminary comparison of the source and sink households of servants.

For this analysis, a sample of servants is followed from birth, to census records of their parental households after their birth, and then on to census records of their service households. By linking the data in this way, we can compare the households that send out young adults as servants to the households that hire those servants.

Data, Methods, and Results

Data

Two primary data sources are used in this study: individual-level census returns, and valuation rolls. These data were collected from the General Registrar of Scotland (GROS) and the Orkney Archive as part of an ongoing study of the population history of the northern Orkney

Islands conducted by the North Orkney Population History Project (Murtha et al. 2008). The decennial censuses provide information on every person at home on the day of enumeration, including age, sex, relationship to the head, and marital status. From these returns, servants are identified, and the composition of the household is obtained. The names of houses and farmsteads, which persist over time, are used to link the census returns to the valuation rolls, which are records of the taxation value of land and buildings. The data used in the first and second parts of this study represent all households on the islands of Westray, Eday, Papa

Westray, Sanday, North Ronaldsay, and Faray for which such record linkage was possible. This sampling method may introduce bias into the data. For instance, the village of Pierowall on

Westray was excluded from the sample because linkage using house names is not possible for the village, as households (particularly compound households) within Pierowall cannot be readily

170 distinguished based on house names alone. In addition, small farmsteads, farmsteads that were abandoned early in the study period, and farmsteads that were infrequently occupied, are less likely to be linked using house names. These farmsteads might not appear in multiple records, the exact name and location of the farmstead may not be clear from historical maps, or the names could have been ―forgotten‖ in the oral history of the islands.

Servants were observed at each decennial census interval from 1851-1901. Servants were identified using the census variable ―relationship to head‖, which indicates that these individuals were ―live-in‖ life-cycle servants, rather than people who lived at home and worked as a ―farm servant‖ for wages, either in a full or part-time capacity.25 Household-level predictor variables were taken from the census record of the household in which the servant was living at the time of the census. Of the total sample population, servants make up a maximum of 8.8 percent in 1851 and a minimum of 4.2 percent in 1901 (Table 4-4). Of the sample population aged 12 to 30, a maximum 28 percent are servants in 1851 and a minimum 13 percent are servants in 1891 (Table 4-5). The sex composition of life-cycle servants is roughly equal, although females usually outnumber males and tend to be 1-2 years older than their male counterparts (Table 4-6). However, the older average age of female servants is perhaps attributable to the tendency for unmarried women to remain in service, and it does not appear to affect women‘s age at marriage in the overall population, which is still younger than men‘s age at marriage, or the age difference between spouses, which also reflects the older average ages of male spouses (Table 4-7). Figures 4-2, 4-3, and 4-4 compare the age distribution of male and female servants with the general sample population. The observed age distribution of servants is

25 In Orkney, ―farm servant,‖ ―agricultural labourer,‖ and ―domestic servant‖ were common occupations (see examples in Chapter 2). However, these occupations were different from life-cycle service, as these individuals received wages for their work, but not room and board. In other words, they lived at home and worked at a nearby farm, perhaps seasonally. These occupations are listed under the census variable ―occupation‖ and their relationship to the head usually indicates that they lived with family.

171 consistent with service as a life-cycle occupation. Note that age-heaping, evident in the distribution of ages in the overall population, is a source of error in the census data.

Household-level Predictors of the Number of Life-cycle Servants: Methods

This part of the study examines which household-level variables are predictive of the number of servants present in the household. Individual-level census returns are used to identify servants and the household-level predictors. Valuation rolls, linked to the census data, provide a proxy measure of landholding size and quality. The predictor variables used include measures of the household, such as C/P ratio, valuation, household type (simple, extended, or compound), and household size.26 Other household measures related to the status of the household head may influence the decision to hire servants. These include the age of the head, sex, marital status, occupation, and if the head is an owner-occupier rather than a tenant. In compound households, head-related variables were calculated using information about the head of the segment of the household in which the servant lived. Therefore, in the case of compound households, only one of the heads listed in the census contribute information to the head-related variables.27 Tables 4-8 through 4-11 present the mean values of these variables for households with at least one servant and households without servants. The sample consists of 2917 total observations of 664 households, as households can be observed over multiple census years. Table 4-12 details the number of households by census year.

26 C/P ratio is calculated using the weights proposed by Hammel (2005). Household types were assigned using the Hammel-Laslett system (1974) and the definition of compound households established in the background section. Servants are not included in the calculation of C/P ratio or household size. 27 Future work will explore other methods to capture the diversity of headship within compound households. Averaging traits across all heads in a compound household may be one way to do this. For example, a household with one male head and one female head would have a value for the sex of the head variable of 0.5 under this strategy.

172 A Poisson regression model is fit to the data, with the number of household servants, including zero, as the outcome variable. The frequency of life-cycle servants and the number of households that include them decrease over the study period, so a linear variable for census year is included to control for the effects of period. Households can appear in the sample in more than one census year, so standard errors are adjusted by clustering over a household identifier.

Household-level Predictors of the Number of Life-cycle Servants: Results

The results of the Poisson regression models are given in Table 4-13. The main effects model and models with household-level and time interaction terms are presented here. The full model (including both household-level and time interactions) has better fit than the main effects

2 model in terms of the Bayesian Information Criterion and the results of a likelihood ratio test (χ (5)

= 161.95, p < 0.001, main effects vs. full model, main effects vs. other interaction models also significant). In the main effects model, terms associated with household-level variables have significant effects on the number of servants. Of these predictors, simple and extended household types are negatively associated with the number of servants relative to compound households.

This finding is likely related to the larger average size of compound households, which have more people, and will therefore have more servants, all other things being equal. Valuation also has significant, but small, positive associations. Interestingly, C/P ratio, which is calculated so that servants are not included, is not a significant predictor. If C/P ratios are calculated so that servants are included, the effect is negative and significant, perhaps indicating that servants may lower household C/P ratios or that some servants were already present when other servants were hired. However, it is important to consider that if servants are included in the measurement of

C/P ratios, it cannot be determined whether servants lower C/P ratios or if households with lower

C/P ratios are better able to afford servant labor. Most variables related to the household head are

173 not significant. However, if the household head is listed in the valuation roll as the owner- occupier, rather than a tenant, then the household tends to have more servants. In addition, if the head is married, the household is likely to have fewer servants than households where the head is single or widowed.

Pairwise interaction terms for valuation, household size, and C/P ratio were added to the main effects model. Prior studies of the formation and dissolution of households found these interaction terms to be important (Jennings et al. 2009a, 2009b). In addition, the interactions of these variables are related to the balance of household labor and land resources, which is likely to influence the demand for servant labor. These three terms were significant and improved model fit. In addition, the numbers of servants in the population, and several other variables, including household size, change over time, so the interaction of several predictor variables and time were tested. The interactions of time and valuation, which is possibly indicative of inflation, and time and household size, were found to be significant (or nearly significant) and model fit was improved.

Once the interactions of household-level variables and time are controlled for, the size of the estimated coefficients of valuation and household size increase and household size reaches levels of statistical significance. The effect of the marital and landowning status of the household head remains largely unchanged. These findings are taken up in detail in the discussion section below.

Household-level Predictors of the Number of Life-cycle Servants by Sex: Methods

This portion of the study compares the number of male and female servants in the household to determine whether household-level variables are associated with the sex composition of the servant workforce. Put another way, these models seek to understand whether

174 the labor of men and women is largely interchangeable, or if household-level characteristics can predict whether a household has additional male or female servants. Here, individual-level census returns are used to determine the number and sex of the servants and identify household- level predictors. As in the previous model, all households are used, so that the number of male or female servants may be zero. Valuation rolls, linked to the census data, provide a proxy measure of landholding size and quality. The predictor variables used include measures of the household, such as C/P ratio, valuation, household type, and household size.28 Other household measures related to the status of the household head may influence the decision to hire servants. These include the age, sex, marital status, and occupation of the head, and whether the head is an owner- occupier or a tenant, all of which are included in the model. An additional variable, the worker sex ratio is added to the sex-specific model.29 This variable accounts for the sex composition of household members of working age (15-65) who are not servants. It is calculated by dividing the total number of males aged 15-65 by the total number of non-servant household members aged

15-65. If the worker sex ratio equals one, then all the non-servant household workers are male, and if the ratio equals zero, then all the non-servant workers are female. The worker sex ratio may determine whether households choose to hire male or female servants in an effort to obtain a desired balance of male and female labor or provide a minimum number of workers of either sex.

Tables 4-14 through 4-17 present the mean values of these variables for households with at least one male servant and households with at least one female servant. The sample consists of 2917 observations of 664 households.

Two Poisson regression models are fit to the data, with the number of female servants and the number of male servants present in the household as the outcome variables. Parameter

28 C/P ratios and household size were calculated after excluding servants. 29 The worker sex ratio variable was tested in the more general number of servants model described above. The coefficient was not significant, and the estimated coefficients for the other variables did not change appreciably.

175 estimates and covariance matrices are obtained from the two Poisson models and entered into a seemingly unrelated regression (SUR) model. This procedure adjusts the estimated standard errors for the simultaneous estimation of the two equations. With these adjustments, it is possible to conduct hypothesis tests of the coefficients across the two equations. For example, it is possible to test whether the estimated coefficient of household size in the equation for the number of male servants is equal to the coefficient of household size in the female servants‘ equation.

The frequency of life-cycle servants and the number of households that include them decrease over the study period, so a linear variable for census year is included to control for the effects of period. Robust standard errors are calculated by clustering over a household identifier. This corrects for the repeated observations of households. Analyses were performed using STATA SE

11 (StataCorp 2009).

Household-level Predictors of the Number of Life-cycle Servants by Sex: Results

The results of the Poisson regression models, adjusted using seemingly unrelated regression, are given in Table 4-18. The main effects model and a model with the same interaction terms as the Poisson model for the number of all servants are presented to facilitate the comparison and interpretation of both portions of the study. A series of Wald tests of linear hypotheses were performed, with p-values adjusted for multiple testing using the Bonferroni method. Here, the equality of estimated coefficients between different equations is tested (H0: βp

[male] = βp [female], where p is the coefficient of interest). The results of these tests are presented in Table 4-19. In the main effects model, the coefficients of valuation, worker sex ratio, the occupation of the head (1 = farmer, 0 = non-farmer), and landowning status (1 = owner- occupier, 0 = tenant) are found to be unequal between the equations for the number of male servants versus female servants. A one-pound increase in valuation increases the expected

176 number of male servants by a greater, although small, amount than the expected number of female servants. Households with a lower worker sex ratio (more females relative to males) are more likely to hire male servants and households with a higher worker sex ratio are more likely to hire female servants. If a household head is a farmer, the expected number of male servants increases, and the expected number of female servants decreases. In addition, the average number of female servants in landowning households is higher than in tenant households.

The same Wald tests were performed for the interaction models. The results are presented in Table 4-19. In these models, the coefficients of valuation, worker sex ratio, the occupation of the head, landowning status, and two of the interaction terms (household size × valuation, and valuation × time), were not equal between the two equations. A one-unit increase in valuation increases the expected number of male servants more than the number of female servants. The coefficient of household sex ratio changes signs between the male and female equations, so that households with a lower worker sex ratio (more females relative to males) are more likely to hire male servants and households with a higher worker sex ratio are more likely to hire female servants. If a household head is a farmer, the expected number of male servants increases, but the expected number of female servants decreases (although the term is not statistically significant at the 0.05 level). Finally, if a household head is an owner-occupier, the expected number of male servants increases (although the size of the increase is small and not statistically significant), but the expected number of female servants increases. In general, the interaction terms have a larger effect on the expected number of male servants.

A Comparison of the Source and Sink Households of Life-cycle Servants

Household C/P ratio is an important predictor of both the number of servants and the sex of servants in the main effects models. While these associations are suggestive of an important

177 connection between the household labor force and the decision to hire servants, these models do not explicitly compare the source and sink households of life-cycle servants. In this portion of the study, servants are identified in census records, and then found in the birth registers. Using the information from the birth records, they were traced back to the census records to find the last time they were observed in a household with at least one of their parents. For each person for which such linkage was possible, a maximum of four census records were included in the sample: the source household at the latest time the index person was observed, the source household for the census interval after that last observation, the sink household where the index individual is first observed, and the sink household in the census interval before the first observation.

The sample consists of 221 individuals who could be linked in this manner, although for some individuals, linkage could not be completed for one or two of the four records described. In other words, to be included in the sample, individuals must be found in both the source and sink households, but sometimes the source household could not be found in next census interval or the sink household could not be found in the preceding census interval. While this sample is very likely to be not representative of the population and is biased toward individuals whose birth record and the census records of both the source and sink households are found in the study area, some trends can be discerned from this dataset. The C/P ratios of the households in the four censuses were calculated, and plotted in histograms (Figures 4-5 through 4-8). These histograms suggest that source and sink households have different distributions of C/P ratios. Source households tend to have more variation in C/P ratios, and slightly higher C/P ratios on average

(Figures 4-5 and 4-7). In addition, source and sink households observed with servants present tend to have slightly higher mean C/P ratios (Table 4-20). In a direct comparison of the source and sink households with the servant present, the source household C/P ratio tends to be higher than the sink household C/P ratio (Figure 4-9). These preliminary and limited results bolster the findings of the multiple regression models that indicate the importance of C/P ratio as a predictor

178 of the hiring of servants. Future work, with a larger linked sample, or with data with more precise time controls (this study is limited to 10-year intervals) may be better able to elucidate the relationships between household labor availability and consumption requirements and the hiring and sending out of servants.

This linked dataset can also be used to address the issue of the relative status of the source and sink households. As noted in the background, there has been some debate in the literature about whether servants circulated among households that were largely social and economic equals, or if servants tended to flow from lower-status households to higher-status households. This limited sample provides evidence in support of the first hypothesis. The majority of source and sink households in this sample were characterized by valuations within

£30 (approximately 1 standard deviation) of each other (Figure 4-10). This preliminary evidence suggests that, at least in Orkney, servants circulated among households of roughly equal economic status, as measured by valuation. Whether this trend is true of other regions, or will hold once a larger linked sample is available, is unknown.

Discussion

The results presented here have implications for the study of life-cycle servants and their importance to the household economy. Several household-level predictors are significantly associated with the number of servants. These include household size, valuation, household type, and interaction terms among valuation, household size, and C/P ratio. In general, larger households and households with higher valuations tend to have higher average numbers of servants, although the sizes of the effects are small. Households headed by a widowed or unmarried person can be expected to have more servants than households headed by married individuals. In addition, landowners, on average, have more servants than tenants, although this

179 term is not significant in the full model, perhaps because households with many servants were rare in the sample. The importance of time and household-level interactions suggests that the relationships among servants, households, and land are complicated and change over time.

Future studies, especially those that consider land use explicitly, may be better able to parse out these relationships.

Interestingly, household C/P ratio was not significantly associated with the number of servants present in the household, while interaction terms that included C/P ratios were significant. In this study, household C/P ratios were calculated after servants were excluded from these measurements. This was done to approximate what the C/P ratio would have been had the household not hired servants and to only include servants on the left side of the regression equation. However, if servants are included in C/P ratios, the coefficients are negative and significant, but difficult to interpret. This is because it cannot be determined whether servants lower C/P ratios or if households with low (more economically favorable) ratios are better able to support servants.

Another interesting negative finding relates to predictor variables of the status of the household head. The age, sex, and occupation of the head were not significantly associated with the number of life-cycle servants present in the household, after controlling for other factors.

These results run somewhat counter to predictions about headship, especially the age of the head, which is usually thought to correspond to stages in household Chayanovian cycles. In addition, one might predict that female-headed households may be facing labor shortages, especially when one considers that men tend to be listed as head rather than women, except in cases where the husband has died. Therefore, the presence of a female head probably indicates that an adult laborer is ―missing‖ from the household. However, it may be that other household members, such as older children and other kin, make up for the lost labor of these missing individuals.

Indeed, there did not seem to be an important ―widowed head‖ effect. An interaction term for

180 unmarried female heads was added to each of the models, but the estimated coefficients of the sex of the head and marital status of the head did not change appreciably and the interaction term was not significant.

The results of the models related to the sex of servants have implications for the debate about the nature of the sexual division of agricultural labor. The coefficients for valuation, worker sex ratio, occupation of the head, landowning status, and several interaction terms were different between the two equations. Households with larger valuations had higher average numbers of servants, but the expected number of male servants increased more than the expected number of female servants for each one-pound increase in valuation. Households appear to hire servants in response to worker sex ratios, as households with higher ratios (more males relative to females) hire more female servants and households with lower ratios hire male servants. Holding other factors constant, non-farming heads were more likely to have higher average numbers of female servants than farming households. The same is true for owner-occupiers, whose average female servant workforce was larger than that of tenants. Perhaps these extra female workers were ―luxury‖ domestic servants rather than agricultural laborers. These female servants may have also worked to prepare food (and perform other domestic tasks) for everyone present in the household, including male servants.

These findings support the premise that male and female labor was not fungible, but rather that the sexes engaged in different forms of labor, although certain tasks may have overlapped. Households may seek a desired balance of male and female workers, as evidenced in the differential effects of household worker sex ratio. Alternately, there may be minimum number of males and females required to operate a holding and households hired servants to meet this minimum. Households with higher valuation, and presumably, more land, hired more men per unit increase in valuation than women. In contrast, owner-occupiers and non-farming households had higher average numbers of female servants. These results suggest that female

181 labor was not exclusively agricultural, as higher average numbers of female servants are associated with non-farming households, where they were likely completing domestic or craft- related tasks. During this period, owner-occupier households were few in number and represented a social middle ground, as they were probably better off than tenant farmers were, but they were not exceptionally wealthy either, as the wealthiest landowners were commonly absentee landlords and their holdings were leased out to smallholders or managed by factors, and were not listed as owner-occupied properties in the historical records. Thus, the finding that owner-occupiers were likely to hire more female servants than tenants is interesting, but difficult to interpret. It is possible that many of these owner-occupier households were those of the clergy or other non-agricultural professionals, such as lighthouse keepers, teachers, doctors, and merchants, who might have had higher demand for female domestic service. However, future study is needed to confirm or refute this speculation.

While this study cannot settle the precise role of life-cycle servants as part of an

―adaptive family economy‖ (Wall 1986), the strength and significance of several household-level and head-related variables as a predictors of both the number and sex of servants demonstrates the important effects that servants can have on the household economy. In addition, the significance of several interaction variables suggests that these relationships are not simple.

Future work in this area might return to the relationships between source and sink households using more completely linked data or using a different data source with better time detail.

Another interesting topic for the Orkney dataset would be the investigation of life-cycle service with respect to the transition to marriage. While the literature focuses upon the function of service that allows for the accumulation of savings before marriage, it is less clear with respect to the social contacts made by servants that may lead to marriage. Indeed, a descriptive study of service and marriage might be informative about the transition to marriage, particularly since it could examine the extent to which servants married family members in the households they

182 worked for, or if they tended to marry people from the islands they moved to for service rather than the islands on which they were born. In addition, servants who marry may also continue to work in a similar capacity (although typically no longer as a ―live-in‖ servant), either full or part- time. Thus, life-cycle service could be studied with regard to how it may lead to employment

(such as wage work) after marriage. Future work might also focus on the relationship between service and illegitimate births. Historical records and ethnographic evidence suggest that illegitimate births were not uncommon in Orkney. Since the majority of life-cycle servants were unmarried and living away from their families, it is possible that births to servants may have formed a significant portion of the total number of non-marital births.

183 Tables and Figures

Table 4-1. Chayanov‘s model of change in household composition (Chayanov 1986: 58).

Figure 4-1. Chayanov‘s model of household cycles in C/P ratio (Chayanov 1986: 59).

184 Table 4-2. Weighting system used to calculate C/P ratios (after Hammel 2005). Production Consumption Male Female Male Female Age Units Age Units Age Units Age Units 0-5 0 0-5 0 0-2 0.1 0-2 0.1 5-7 0.1 5-6 0.2 2-5 0.3 2-5 0.3 7-9 0.2 6-10 0.5 5-9 0.5 5-6 0.5 9-12 0.5 10-15 0.7 9-12 0.7 6-10 0.7 12-15 0.9 15-20 0.7 12-15 0.8 10-12 0.8 15-50 1 20-60 0.8 15-50 1 12-60 0.8 50+ 0.8 60+ 0.7 50+ 0.8 60+ 0.7

Table 4-3. Birthplaces of servants, 1851-1901. All servants are observed in the study area. Caithness is the mainland Scottish county nearest to Orkney. Shetland is an archipelago north of Orkney. Males Females Study Area 731 800 Other Orkney 48 71 Shetland and Caithness 16 38 Other Scotland 14 6 Outside of Scotland 3 3 Missing 1 3 Total 813 921

Table 4-4. Total sample population by census year. This sample is less than the total population of the islands (Table 1-1) because of sample selection criteria that include only households that can be linked using house names. Year Non-Servants Servants Total Percent Servants 1851 4646 409 5055 8.80 1861 4902 351 5253 7.16 1871 4957 334 5291 6.74 1881 5099 250 5349 4.90 1891 4611 197 4808 4.27 1901 4222 178 4400 4.22

Table 4-5. Sample population between ages 12 and 30, by census year.

Year Non-Servants Servants Percent Servants 1851 1261 351 27.84 1861 1325 298 22.49 1871 1308 284 21.71 1881 1449 214 14.77 1891 1290 173 13.41 1901 1105 155 14.03

185 Table 4-6. Number and mean age of servants, by sex and census year.

Census Year Males Mean Age of Males Females Mean Age of Females 1851 207 19.42 202 22.99 1861 164 20.89 187 22.77 1871 150 20.97 184 22.47 1881 112 20.97 138 23.39 1891 87 21.23 110 22.19 1901 91 21.60 87 22.52

Table 4-7. Distribution of individual age differences between spouses, first marriages only (includes both servants and non-servants).

Mean Age Marriage Number of Difference Year Marriages (M-F) Std. Dev. 1851-1860 189 1.49 5.74 1861-1870 313 2.69 5.83 1871-1880 274 2.32 5.43 1881-1890 248 2.52 5.08 1891-1900 219 2.34 4.92

186 Figure 4-2. Histograms of the age distribution of male servants, by census year.

1851 1861 1871

60

40

20 0

1881 1891 1901

60

Frequency

40

20 0 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 Age of Male Servants

Figure 4-3. Histograms of the age distribution of female servants, by census year.

1851 1861 1871

60

40

20 0

1881 1891 1901

60

Frequency

40

20 0 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 Age of Female Servants

187 Figure 4-4. Histograms of the age distribution of the sample population by census year.

1851 1861 1871

400

300

200

100 0

1881 1891 1901

400

Frequency

300

200

100 0 0 50 100 0 50 100 0 50 100 Age of Non-Servants

Table 4-8. Mean values of the household-level predictor variables for households with no servants. The presence of servants is not inconsistent with simple households as defined here. Mean Proportion of Mean Mean CP Mean Simple Household Census year ratio Valuation households Size N 1851 1.11 6.93 0.45 6.95 432 1861 1.12 8.05 0.41 7.14 469 1871 1.11 8.25 0.46 6.97 487 1881 1.09 10.17 0.41 6.78 545 1891 1.09 10.55 0.47 6.46 534 1901 1.07 9.30 0.48 6.11 544

188 Table 4-9. Mean values of the household-level predictor variables for households with servants. The presence of servants is not inconsistent with simple households as defined here. Mean Proportion of Mean Mean CP Mean Simple Household Census year ratio Valuation households Size N 1851 1.08 40.56 0.30 11.90 167 1861 1.10 44.01 0.29 11.96 157 1871 1.09 47.09 0.28 12.17 163 1881 1.09 69.01 0.31 12.12 140 1891 1.08 60.86 0.37 12.18 117 1901 1.07 56.42 0.42 11.28 104

Table 4-10. Mean values of head-related predictor variables for households with no servants.

Proportion of Proportion of Proportion of currently farming owner- Mean age of Proportion of married household occupier Census Year head female heads heads heads heads N 1851 54.86 0.24 0.79 0.76 0.02 432 1861 54.19 0.20 0.78 0.76 0.02 469 1871 56.04 0.21 0.56 0.78 0.02 487 1881 56.03 0.20 0.78 0.78 0.02 545 1891 56.69 0.19 0.78 0.76 0.01 534 1901 58.03 0.16 0.75 0.82 0.01 544

Table 4-11. Mean values of head-related predictor variables for households with servants.

Proportion of Proportion of Proportion of currently farming owner- Mean age of Proportion of married household occupier Census Year head female heads heads heads heads N 1851 58.36 0.21 0.65 0.83 0.09 167 1861 58.52 0.22 0.68 0.85 0.10 157 1871 57.77 0.21 0.52 0.85 0.06 163 1881 56.90 0.18 0.67 0.88 0.04 140 1891 59.45 0.23 0.65 0.78 0.03 117 1901 57.57 0.22 0.56 0.85 0.04 104

189 Table 4-12. The number of households in the sample by census year and presence of servants.

Households Households with at Census without least 1 year servants servant 1851 432 167 1861 469 157 1871 487 163 1881 545 140 1891 534 117 1901 544 104

Table 4-13. Results of Poisson regression models for the number of servants present in the household (N=664 households, with 2917 total observations, as households can be present in more than one census year). Standard errors are adjusted for multiple observations of households by clustering over a household identifier.

Main HH Time Effects Interactions Interactions Full Model Household-level variables CP ratio 0.1295 0.1278 0.1924 0.2383 Valuation 0.0098 *** 0.0112 *** 0.0117 *** 0.0132 *** Simple Household -0.4954 *** -0.3758 ** -0.4902 ** -0.3761 ** Extended Household -0.6000 ** -0.5412 ** -0.5979 *** -0.5393 ** Compound Household ref. ref. ref. ref. Household size, excluding servants if present -0.0126 0.0214 *** -0.0015 0.0255 ** Household-level interactions Valuation X Household size -0.0205 ** -0.0210 *** Household size X CP ratio 0.2634 *** 0.2505 *** Valuation X CP ratio -0.0002 *** -0.0002 *** Variables related to household head Age of the head 0.0026 0.0016 0.0033 0.0020 Sex of head (1=female, 0=male) -0.1654 -0.1549 -0.1575 -0.1423 Marital status of head (1=married, 0=single or widowed) -0.3075 ** -0.2758 ** -0.2943 ** -0.2590 ** Occupation of Head (1=farm, 0=non-farm) 0.1337 0.0975 0.1166 0.0841 Land owner (1=yes, 0=no) 0.4479 * 0.4805 ** 0.3909 + 0.4522 * Time Interactions Household size X Time -0.0042 -0.0006 Valuation X Time -0.0005 -0.0018 Time Controls Time -0.2425 *** -0.2256 *** -0.2107 -0.1943 *** Intercept 0.3417 + 0.1568 0.2379 0.0532

BIC 5223.2170 5099.7720 5214.7000 5101.1610 R-squared 0.2197 0.2421 0.2234 0.2443 +p<.1; *p<.05; **p<.01; ***p<.001

190 Table 4-14. Mean values of household-level predictor variables for households with at least one male servant. Proportion of Mean Census Mean Simple Household Worker Sex Year Mean CP Valuation Households Size Ratio 1851 1.07 50.33 0.27 13.19 0.49 1861 1.09 64.09 0.23 13.22 0.46 1871 1.08 63.56 0.23 13.27 0.47 1881 1.07 62.14 0.27 11.30 0.42 1891 1.07 66.02 0.30 11.83 0.44 1901 1.06 49.81 0.42 8.37 0.47

Table 4-15. Mean values of household-level predictor variables for households with at least one female servant. Proportion of Mean Census Mean Simple Household Worker Sex Year Mean CP Valuation Households Size Ratio 1851 1.09 44.36 0.31 12.68 0.54 1861 1.10 48.80 0.31 12.17 0.53 1871 1.09 52.34 0.28 11.40 0.53 1881 1.09 84.29 0.35 11.55 0.50 1891 1.09 70.02 0.39 11.04 0.54 1901 1.07 65.29 0.44 10.71 0.54

Table 4-16. Mean values of head-related predictor variables for households with at least one male servant. Proportion of Proportion of Proportion of Census Mean Age of Proportion of Currently Farming Landowning Year Head Female Heads Married Heads Heads Heads 1851 60.63 0.23 0.85 0.94 0.06 1861 58.67 0.21 0.85 0.94 0.06 1871 57.16 0.23 0.71 0.92 0.06 1881 57.19 0.14 0.84 0.93 0.03 1891 60.27 0.22 0.81 0.95 0.00 1901 56.95 0.15 0.71 0.95 0.03

Table 4-17. Mean values of head-related predictor variables for households with at least one female servant. Proportion of Proportion of Census Mean Age of Proportion of Currently Proportion of Landowning Year Head Female Heads Married Heads Farming Heads Heads 1851 58.21 0.17 0.85 0.81 0.09 1861 58.37 0.23 0.87 0.82 0.12 1871 57.53 0.19 0.57 0.83 0.05 1881 56.35 0.16 0.87 0.85 0.06 1891 57.45 0.20 0.84 0.71 0.04 1901 56.38 0.25 0.73 0.81 0.03

191 Table 4-18. Results of the Poisson regression models for the number of servants, by sex. Standard Errors adjusted for simultaneous estimation using seemingly unrelated regression and multiple observations of households by clustering over a household identifier. N=664 households, with 2917 total observations, as households can be present in more than one census year. Main Main Full Full Effects, Effects, Model, Model, Male Female Male Female Household-level variables CP ratio 0.1713 0.4874 0.2239 0.6302 Valuation 0.0105 *** 0.0089 *** 0.0159 *** 0.0104 *** Simple Household -0.4936 ** -0.5536 *** -0.3375 + -0.4708 ** Extended Household -0.5391 * -0.6701 *** -0.4509 * -0.6352 *** Compound Houseold ref. ref. ref. ref. Household size, excluding servants if present -0.0241 + -0.0030 0.0391 *** 0.0242 ** Worker Sex Ratio -0.5478 1.7278 *** -0.7012 + 1.6889 *** Household-level interactions Valuation X CP ratio -0.0166 * -0.0229 *** Household Size X CP ratio 0.3033 ** 0.1857 * Valuation X Household size -0.0003 *** -0.0001 ** Variables related to household head Age of the head 0.0009 0.0037 0.0000 0.0028 Sex of head (1=female, 0=male) -0.1501 -0.0658 -0.0978 -0.0669 Occupation of Head (1=farm, 0=non-farm) 0.7110 ** -0.1999 0.6502 * -0.2197 Marital status of head (1=married, 0=single or widowed) -0.3310 * -0.2351 * -0.2623 * -0.1968 + Land owner (1=yes, 0=no) 0.0960 0.6831 *** 0.0901 0.7165 *** Time Interactions Valuation X Time -0.0011 *** -0.0001 Household size X Time -0.0062 * -0.0015 Time Controls Time -0.2618 *** -0.2153 *** -0.1774 *** -0.1940 *** Intercept -0.6262 -1.2613 *** -0.9614 * -1.3742 ***

R-squared 0.1953 0.2048 0.2289 0.2181 +p<.1; *p<.05; **p<.01; ***p<.001

192 Table 4-19. Results of Wald tests of the equality of coefficients across equations.

Main Effects Full Model

Chi-square, Chi-square, Coefficent 1 df p-value* 1 df p-value* CP ratio 0.30 1.0000 0.47 1.0000 Valuation 8.57 0.0411 15.21 0.0016 Simple Household (Compound=reference) 0.14 1.0000 0.61 1.0000 Extended Household (Compound=reference) 0.75 1.0000 1.56 1.0000 Household size, excluding servants if present 4.42 0.4265 3.64 0.9581 Worker Sex Ratio 28.83 <.00001 29.01 <.00001 Age of the head 0.39 1.0000 0.39 1.0000 Sex of head (1=female, 0=male) 0.40 1.0000 0.04 1.0000 Occupation of Head (1=farm, 0=non-farm) 13.08 0.0036 11.31 0.0131 Marital status of head (1=married, 0=single or widowed) 0.42 1.0000 0.20 1.0000 Land owner (1=yes, 0=no) 9.43 0.0256 9.16 0.0421 Valuation X CP ratio 0.55 1.0000 Household size X CP ratio 0.93 1.0000 Valuation X Household size 10.04 0.0260 Valuation X Time 9.05 0.0448 Household size X Time 0.80 1.0000 Time 1.87 1.0000 0.28 1.0000 *Adjusted for multiple testing using Bonferroni Method

193 Figure 4-5. Histogram of the C/P ratio of the source household when the future servant was last

observed.

8

7

6

5

4

Density

3

2

1 0 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 CP ratio

Figure 4-6. Histogram of the C/P ratio of the sink household in the census interval before the

servant arrives.

8

7

6

5

4

Density

3

2

1 0 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 CP ratio

194 Figure 4-7. Histogram of the C/P ratio of the source household in the census interval after the

servant leaves.

8

7

6

5

4

Density

3

2

1 0 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 CP ratio

Figure 4-8. Histogram of the C/P ratio of the sink household with the servant observed in it.

8

7

6

5

4

Density

3

2

1 0 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 CP ratio

195 Table 4-20. Distribution of household C/P ratio in the four census records.

N Mean Std. Dev. Min Max Sink before servant 192 1.0797 0.0687 0.9706 1.3407 Sink with servant 213 1.0802 0.0765 0.9706 1.4063 Source with servant 215 1.1708 0.1010 1.0000 1.6667 Source after servant 175 1.1055 0.0970 0.9714 1.5000

Figure 4-9. Distribution of the difference in C/P ratio between source and sink households (both

with the servant present). Mean=.087, Standard Deviation=.126

4

3

2

Density

1 0 -.4 -.2 0 .2 .4 .6 Source CP ratio - Sink CP ratio

Figure 4-10. Distribution of the difference in valuation between source and sink households

(both with the servant present). Mean=4.32, Standard Deviation=29.15

.025

.02

.015

Density

.01

.005 0 -200 -100 0 100 200 Source Valuation - Sink Valuation

196 References

Anderson, M. 1971. Family Structure in Nineteenth Century Lancashire. Cambridge: Cambridge University Press.

Arru, A. 1990. "The distinguishing features of domestic service in Italy." Journal of Family History 15(4):547-566.

Blackstone, W. 1979. Commentaries on the Laws of England: A Facsimile of the First Edition of 1765-1769. Chicago: University of Chicago Press.

Buchan, D. 1984. "The expressive culture of nineteenth-century Scottish farm servants." Pp. 226- 242 in Farm Servants and Labour in Lowland Scotland, 1770-1914, edited by T.M. Devine. Edinburgh: John Donald.

Campbell, C. and J.Z. Lee. 1996. "A death in the family: Household structure and mortality in rural Liaoning: Life-event and time-series analysis, 1792–1867." The History of the Family 1(3):297-328.

Chayanov, A.V. 1986. The Theory of Peasant Economy. Madison: University of Wisconsin Press.

Chibnik, M. 1984. "A cross-cultural examination of Chayanov's theory." Current Anthropology 25(3):335.

Cooper, S.M.I. 2004. "From family member to employee: Aspects of continuity and discontinuity in English domestic service, 1600-2000." Pp. 277-296 in Domestic Service and the Foundation of European Identity, edited by A. Fauve-Chamoux. New York: Peter Lang.

—. 2005. "Service to servitude? The decline and demise of life-cycle service in England." The History of the Family 10(4):367-386.

Devine, T.M. 1984a. "Introduction: Scottish farm service in the Agricultural Revolution." Pp. 1-9 in Farm Servants and Labour in Lowland Scotland, 1770-1914, edited by T.M. Devine. Edinburgh: John Donald.

—. 1984b. "Women workers, 1850-1914." Pp. 98-123 in Farm Servants and Labour in Lowland Scotland, 1770-1914, edited by T.M. Devine. Edinburgh: John Donald.

Dribe, M. 2000. Leaving Home in a Peasant Society: Economic Fluctuations, Household Dynamics and Youth Migration in Southern Sweden, 1829 - 1866. Sodertalje: Almqvist & Wiksell International.

Engelen, T. and A.P. Wolf. 2005. "Introduction: Marriage and the family in Eurasia. Perspectives on the Hajnal hypothesis." Pp. 15-36 in Marriage and the Family in Eurasia: Perspectives on the Hajnal Hypothesis, edited by T. Engelen and A.P. Wolf. Amsterdam: Aksant.

197 Ewan, E. 2004. "Mistresses of themselves? Female domestic servants and by-employments in sixteenth century Scottish towns." Pp. 411-434 in Domestic Service and the Formation of European Identity, edited by A. Fauve-Chamoux. New York: Peter Lang.

Fauve-Chamoux, A. and R. Wall. 2005. "Domestic servants in comparative perspective: Introduction." The History of the Family 10:345-354.

Fenton, A. 1976. Scottish Country Life. Edinburgh: John Donald.

—. 1997. The Northern Isles: Orkney and Shetland. East Linton: Tuckwell Press.

Flinn, M.W., J. Gillespie, N. Hill, A. Maxwell, R. Mitchison, and C. Smout. 1977. Scottish Population History: From the 17th Century to the 1930s. Cambridge: Cambridge University Press.

Goldberg, P.J.P. 1992. Women, Work, and Life Cycle in a Medieval Economy: Women in York and Yorkshire c. 1300-1520. New York: Oxford University Press, USA.

Gray, J. 2006. "Gender composition and household labour strategies in pre-Famine Ireland." The History of the Family 11(1):1-18.

Gray, M. 1984. "Farm workers in North-east Scotland." Pp. 10-28 in Farm Servants and Labour in Lowland Scotland, edited by T.M. Devine. Edinburgh: John Donald.

Hagen, E.H., H.C. Barrett, and M.E. Price. 2006. "Do human parents face a quantity-quality tradeoff?: Evidence from a Shuar community." American Journal of Physical Anthropology 130(3):405-418.

Hajnal, J. 1953. "Age at marriage and proportions marrying." Population Studies 7(2):111-136.

—. 1965. "European marriage patterns in perspective." Pp. 101-143 in Population in History, edited by D.V. Glass and D.E.C. Eversley. London: Arnold.

—. 1982. "Two kinds of preindustrial household formation systems." Population and Development Review 8(3):449-494.

Hammel, E.A. 2005. "Chayanov revisited: A model for the economics of complex kin units." Proceedings of the National Academy of Sciences 102(19):7043-7046.

Hammel, E.A. and P. Laslett. 1974. "Comparing household structure over time and between cultures." Comparative Studies in Society and History 16(1):73-109.

Harrison, W. 1968. The Description of England. Ithaca: Cornell University Press.

Hasbach, W. 1908. A History of the English Agricultural Labourer. London: PS King & Son.

Howatson, W. 1984. "Grain Harvesting and Harvesters." Pp. 124-142 in Farm Servants and Labour in Lowland Scotland, 1770-1914, edited by T.M. Devine. Edinburgh: John Donald.

198 Hunt, D. 1979. "Chayanov's model of peasant household resource allocation." Journal of Peasant Studies 6(3):247-285.

Jennings, J.A., C.S. Sparks, J.W. Wood, P.L. Johnson, T.M. Murtha, and S.A. Matthews. 2009a. "Dissolution of extended-family households in Northern Orkney, Scotland, 1851-1901." Presented at the annual meeting of the Population Association of America, Detroit, MI.

—. 2009b. "Formation and dissolution of extended-family households in Orkney, Scotland, 1851- 1901." Presented at the meeting of the International Union for the Scientific Study of Population, Marrakech, Morocco.

Jennings, J.A. and J.W. Wood. 2009. "Measuring Consumer-to-Producer Ratios: Do Weighting Systems Matter?" Presented at the annual meeting of the Social Science History Association, Long Beach, CA.

Kertzer, D.I. 1984. Family Life in Central Italy, 1880-1910: Sharecropping, Wage Labor, and Coresidence. New Brunswick: Rutgers University Press.

Kertzer, D.I. and M. Barbagli. 2002. The History of the European Family: Family life in the long nineteenth century, 1789-1913. New Haven: Yale University Press.

Kertzer, D.I. and D.P. Hogan. 1990. "Household organization and migration in nineteenth- century Italy." Social Science History 14(4):483-505.

Kitchen, F. 1981. Brother to the Ox: The Autobiography of a Farm Labourer. Horsham, Sussex: Caliban Books.

Kussmaul, A. 1981a. Servants in Husbandry in Early Modern England. Cambridge: Cambridge University Press.

Kussmaul, A.S. 1981b. "The ambiguous mobility of farm servants." The Economic History Review 34(2):222-235.

Landsteiner, E. 1999. "Household, family, and economy among wine-growing peasants The case of lower Austria in the first half of the nineteenth century." The History of the Family 4(2):113-135.

Laslett, P. 1977. Family life and illicit love in earlier generations: Essays in historical sociology. New York: Cambridge University Press.

—. 1988. "Family, kinship and collectivity as systems of support in pre-industrial Europe: A consideration of the ‗nuclear-hardship‘ hypothesis." Continuity and Change 3(2):153- 175.

—. 2000. The World We Have Lost: Further Explored. London: Routledge.

Levitt, I. and C. Smout. 1984. "Farm worker's incomes in 1843." Pp. 156-187 in Farm Servants and Labour in Lowland Scotland, 1770-1914, edited by T.M. Devine. Edinburgh: John Donald.

199 Lewis, J.V.D. 1981. "Domestic labor intensity and the incorporation of Malian peasant farmers into localized descent groups." American Ethnologist 8(1):53-73.

Molin, G.D. 1990. "Family forms and domestic service in Southern Italy from the seventeenth to the nineteenth centuries." Journal of Family History 15(4):503-527.

Murtha, T.M., P.L. Johnson, J.W. Wood, S.A. Matthews, C.S. Sparks, and J.A. Jennings. 2008. "Historical demography, oral history, settlement archaeology, and landscape ecology: The North Orkney Population History Project." Presented at the annual meeting of the European Population Conference, Barcelona, Spain.

Nagata, M.L. 2005. "One of the family: Domestic service in early modern Japan." The History of the Family 10(4):355-365.

Netting, R.M. 1993. Smallholders, Householders: Farm Families and the Ecology of Intensive, Sustainable Agriculture. Stanford: Stanford University Press.

Orr, A. 1984. "Farm servants and farm labour in the Forth Valley and South-east Lowlands." Pp. 29-54 in Farm Servants and Labour in Lowland Scotland, 1770-1914, edited by T.M. Devine. Edinburgh: John Donald.

Palsson, H. and P. Edwards. 1981. Orkneyinga Saga: The History of the Earls of Orkney. London: Penguin Books.

Perz, S.G. 2003. "Household demography and land use allocation among small farms in the Brazilian Amazon." Human Ecology Review 9(2):1-16.

Pinchbeck, I. 1969. Women Workers and the Industrial Revolution, 1750-1850. London: Virago.

Redfield, R. 1989. The Little Community and Peasant Society and Culture. Chicago, IL: University of Chicago Press.

Reher, D.S. 1998. "Family ties in Western Europe: persistent contrasts." Population and Development Review 24(2):203-234.

Shaw, F.J. 1980. The Northern and Western Islands of Scotland: Their economy and society in the seventeenth century. Edinburgh: John Donald.

Smith, R.M. 1984. "Families and their property in rural England, 1250-1800." Pp. 1-86 in Land, Kinship and Life-cycle, edited by R.M. Smith. Cambridge: Cambridge University Press.

Smout, T.C. 1986. A Century of the Scottish People, 1830-1950. New Haven: Yale University Press.

Snell, K.D.M. 1985. Annals of the Labouring Poor: Social Change and Agrarian England, 1660- 1900. Cambridge: Cambridge University Press.

Sparks, C.S. 2007. Households, Land, and Labor: Population Dynamics in the Northern Orkney Islands, Scotland, 1851 to 2003. Ph.D. Dissertation, Anthropology and Population Research Institute, Pennsylvania State University.

200 StataCorp. 2009. Stata Statistical Software, Release 11. College Station, TX: StataCorp LP.

Stone, L. 1979. The Family, Sex and Marriage in England 1500-1800. New York: Penguin.

Thomson, W.P.L. 2008a. The New History of Orkney. Edinburgh: Birlinn Ltd.

—. 2008b. Orkney Land and People. Kirkwall: The Orcadian Limited (Kirkwall Press).

Viazzo, P.P., M. Aime, and S. Allovio. 2005. "Crossing the boundary: Peasants, shepherds, and servants in a western Alpine community." The History of the Family 10(4):387-405.

Wall, R. 1986. "Work, welfare, and the family: An illustration of the adaptive family economy." Pp. 261-294 in The World We Have Gained: Histories of Population and Social Structure, edited by L. Bonfield, R.M. Smith, and K. Wrightson. New York: Basil Blackwell.

Watson, J. 1894. The Annals of a Quiet Valley, by a Country Parson. London: J. M. Dent.

Whittle, J. 2000. The Development of Agrarian Capitalism: Land and labour in Norfolk, 1440- 1580. New York: Oxford University Press, USA.

—. 2005. "Servants in rural England c. 1450-1650: hired work as a means of accumulating wealth and skills before marriage." Pp. 89-110 in The Marital Economy in Scandinavia and Britain, 1400-1900, edited by M. Agren and A.L. Erickson. Burlington, VT: Ashgate.

Whyte, I.D. 1989. "Population mobility in early modern Scotland." Pp. 37-58 in Scottish Society, 1500-1800, edited by R.A. Houston and I.D. Whyte. Cambridge: Cambridge University Press.

Whyte, I.D. and K.A. Whyte. 1988. "The geographical mobility of women in early modern Scotland." Pp. 83-106 in Perspectives in Scottish Social History: Essays in Honour of Rosalind Mitchison, edited by L. Leneman. Aberdeen: Aberdeen University Press.

Wolf, E.R. 1966. Peasants. Englewood Cliffs: Prentice-Hall.

Chapter 5

Discussion and Future Directions

Like other preindustrial farming households, households in Orkney were the primary units of production and consumption. Household members contributed to subsistence production, often beginning at young ages, and obtained food, shelter, and other necessities from the pooled product of household labor. Households, therefore, organized much of the interaction of individuals with the environment, most notably land resources. The choice of the household as a unit of analysis also allows for the study of behavior at a level of aggregation above the individual but below the population. In addition, activities important to both economic and ecological studies, such as the sharing of resources and preparation and consumption of food, take place in households. Since people are grouped into households when undertaking essential subsistence activities and making decisions with economic and demographic consequences, households and how they vary in size, age, sex, and kin composition, and other characteristics are essential to the study of the agrarian population dynamics.

Pioneering studies of Northwest European historical demography discovered that, contrary to popular notions, households in the past were small (usually 5-7 people on average) and nuclear in organization (Anderson 1995; Laslett 2000; Laslett et al. 1983; Laslett and Wall

1974). However, later studies demonstrated significant regional variation in household forms within Europe (Berkner 1972; Czap 1983; Engelen and Wolf 2005; Erdozáin-Azpilicueta and

Mikelarena-Peña 1998; Szoltysek 2008). Evidence from this thesis demonstrates that household organization in late nineteenth century Orkney is more complex than the general Northwest

European pattern of predominantly small simple-family households would lead us to expect.

Many essential features of the Northern Orkney household (such as the existence of compound

202 households) were clarified using multiple data sources, including archaeological information, which earlier historical demographers did not have at their disposal. Perhaps other regions of

Northwest Europe feature more complex household forms than the documentary evidence, such as census listings, alone would lead us to believe. If new datasets are developed for other areas of

Northwest Europe (such as England), some previously ―hidden‖ forms of household complexity may be revealed. It is essential to reevaluate our understanding of preindustrial households as new data sources become available, and this collection of papers represents an attempt to understand household membership, dynamics, and economics in light of new information.

Discussion

Chapter 2, ―Households in Northern Orkney, Scotland, 1851-1901,‖ demonstrates how historical documentary evidence (censuses, vital registers, and valuation rolls) can be combined with archaeological surveys and historical maps to enhance understanding of household form and function by linking households to the physical structures they inhabited. This use of combined data sources is uncommon in historical demography and augments customary interpretations of written sources. In fact, a distinct type of household, the compound household, that can be identified using multiple data sources, would otherwise have gone unnoticed if only census listings were used to determine household membership. The three elements of the definition of a compound household, a farmstead with multiple dwellings, a single shared set of farm buildings, and multiple heads listed in the census, have been found in each of the cases examined to date where sufficient archaeological evidence has allowed for the identification of building types. In

Northern Orkney, and perhaps elsewhere, the relationship between this type of census listing and farmstead structure appears to be consistent enough to justify extending the definition to census

203 listings with multiple heads at a single farmstead to cases where supporting archaeological evidence is not available.

By this definition of compound households, the distribution of household types in late nineteenth century Orkney is different from that found in England, where nuclear (simple) households are common (Laslett 2000). However, it may be the case that more complex living arrangements are ―hidden‖ in the census data from England as well. If census listings can be combined with other information about the living arrangements in English rural communities, then these studies, and their conclusions about the dominance of the nuclear family household in the past, should be reexamined. Orkney‘s high frequency of complex households, the category into which compound households are grouped, is much more like that found in Southern and

Eastern Europe (Berkner 1972; Czap 1982; Kertzer 1984; Wall 2001). Why does the frequency of household complexity in Orkney break with the pattern broadly described for Northwest

Europe (Engelen and Wolf 2005; Hajnal 1982; Laslett et al. 1983; Laslett and Wall 1974)? There are several competing explanations for this pattern. The first is the simplest. When any generalization is made about a broad geographic area, it is probable that at least some local studies will find different patterns than those that characterize the region as a whole. However, this explanation does not offer suggestions about why Orkney would be the local area that breaks with regional trends.

Orkney is different from the rest of Northwest Europe in several ways. Orcadians appear to have practiced impartible inheritance, wherein a farmstead and its holdings are passed intact from one generation to the next. This form of inheritance usually implies that one child is designated as an heir, a characteristic of stem household systems. However, stem-family households are only complex households for a few years in the household life-cycle, when three generations are alive and simultaneously coresident. Thus, it is unlikely that this practice raises the proportion of complex households to the high level observed in Orkney. Another contributing

204 factor may be the scarcity of land imposed by this well-populated island environment.

Geography, therefore, may reduce the ability of children to find a farmstead of their own at the time of marriage. A compound household, which offers more privacy and living space than a non-compound, but otherwise complex, household (i.e. an extended- or multiple-family occupying a single dwelling), may offer a solution to this dilemma. Out-migration is another solution to the problem of farmstead shortage, and records indicate that it was a common practice and may have been becoming more common over the historical period examined in these studies.

Finally, the mixed economy of the islands (farming, animal husbandry, fishing, kelp-making, craft production, etc.) may have also contributed to household complexity. Compound households, with their larger size, could have allowed for greater diversity in the household labor force and thus a wider range of household economic activities than was feasible in a simple household, all other things (landholding size, etc) being equal.

Chapter 3, ―The formation and dissolution of extended-family and compound households in Northern Orkney, Scotland, 1851-1901,‖ explores further the differences between compound and non-compound extended households by investigating the timing of transitions between non- compound and compound household forms and extended and simple household forms. Internal household dynamics, such as Chayanovian cycles, may operate differently in these household forms (Chayanov 1986; Hammel 2005). The effects of covariates on the formation and dissolution of compound and extended households are found to be different, reinforcing the idea that compound households are fundamentally different from extended ones. The differential effects of the number of births and deaths in the household are particularly striking. The number of deaths in the household are an important predictor of the dissolution of extended households, but not compound households. This finding is probably related to the smaller size of extended households. A household may be extended by the presence of only one additional relative, such as an elderly, widowed parent. If that one member dies, the household immediately changes to a

205 simple form. In contrast, births are an important predictor of the formation of compound households, but not extended households. One possible interpretation of this finding is that an expanding household may seek to add a dwelling to the farmstead to accommodate new members. The household C/P ratio, a measure of household composition using age- and sex- based production and consumption weights, is an important predictor for both household types.

However, as discussed in Chapter 3, its interpretation can be problematic. Yet, it is possible to conclude that household composition, summarized by the C/P ratio, has an important effect on household dynamics. Parsing out the exact nature of that effect, and how it varies by household type, remains a subject for future study.

Chapter 4, ―Household-level predictors of the presence of life-cycle servants in Northern

Orkney, Scotland, 1851-1901,‖ highlights one way that households may alter their composition independently of their demographic fortunes: by hiring life-cycle servants. This chapter also examines how households operate as adaptive economic entities that, at times, can adjust their labor force to suit particular needs. Here again, C/P ratio was an important predictor variable, this time in relation to the number of servants present in the household. Household size and valuation were positively and significantly associated with the number of servants, suggesting a relationship between the size of the household labor force and size of landholdings. In addition, the numbers of male and female servants were compared to explore whether servants were hired differentially by sex or if their labor was largely interchangeable. The results of the analysis suggest that male and female labor was not identical. Rather, households appear to hire males or females differently by the sex composition of the adults of working age present in the household.

This implies that households sought to balance the ratio of male and female workers in the household, or at least obtain a minimal threshold of laborers of each sex. This result makes sense in light of anecdotal historical evidence on the division of labor by sex in traditional Scottish farming.

206 Future Directions

A common theme of the household unites these three studies. Chapter 2 describes households using multiple data sources, while Chapter 3 explores determinants of household dynamics and Chapter 4 demonstrates the economic adaptability of households through the hiring of life-cycle servants and the sexual division of labor within the household. While households are an important unit of analysis, especially in traditional farming societies where they are usually coterminous with the effective farming unit, they are also the context for individual action.

Future work using this dataset should address the role of the individual within the context of his or her household. Unfortunately, this task is difficult at present because of incomplete data linkage. Indeed, data linkage is a priority for future work. For instance, Chapter 3 could be improved greatly with a larger sample size, which was limited in this study by the incompleteness of linkage. While record linkage is time-consuming and tedious, it is certainly worthwhile insofar as it will allow for improved prospective analysis of individual life courses.

The North Orkney Population History dataset is a rich resource and this thesis represents one early attempt at analysis. Several additional data sources exist that were not directly employed here, especially information about land quality, land use change, ethnographic interviews, and time series of staple grain prices. Surely future work will be enhanced by the inclusion of these resources. Another area for future work involves the development of measures of household composition. This study used household C/P ratios, a variant of household dependency ratios, but their interpretation was difficult. Part of this difficultly was related to the relatively long intervals (ten years) between observations of household composition in the decennial censuses that are inherent in the data. However, a better measure, or set of measures, of household composition may improve the study of the effects of household composition on demographic outcomes.

207 Comparative work will also improve understanding of some of the issues raised in this thesis. For example, multiple island populations could be compared to continental populations in order to explore the possible effects of the restricted land availability and high costs of off-island travel often associated with island environments. Comparative work within Orkney is another area for future research. This thesis treated the six islands of the study area as if they were the same (statistical analyses were not stratified by island, for example). However, there is environmental (and presumably land use) variation among these islands, as well as variation in migration patterns, especially in-migration in more recent years. The comparative study of these islands may inform research related to the differential effects of geography on demographic outcomes.

A final set of questions that needs to be addressed in future research has to do with the changes in household structure and dynamics over the second half of the nineteenth century revealed by these analyses. These decades represented a period of change in Orkney. They saw the end of agricultural expansion and the beginning of agricultural and population decline.

Households in this period began to hire fewer servants (Chapter 4). The proportion of compound households decreased, while the proportion of extended and simple households increased

(Chapter 2). These changes were reflected in the decreasing hazard of compound household formation over the period of interest (Chapter 3). However, these changes have not yet been studied in conjunction with the social and economic factors that were also changing during this period. Future work should attempt to connect changes in household dynamics and structure with these other forces. This might be accomplished through studies of the effects of land use change

(such as farm consolidation and improvement) on household structure. The potential effect of changes in staple grain prices on household dynamics and demographic outcomes, such as increased mortality or out-migration in response to unfavorable prices, is also of interest. In addition, parsing out the exact nature of part-time, seasonal, and bye-employments may also shed

208 light on alternatives to direct participation in the household‘s farming enterprise and how that may, in turn, influence household and population dynamics.

209

References

Anderson, M. 1995. Approaches to the History of the Western Family, 1500-1914. Cambridge: Cambridge University Press.

Berkner, L.K. 1972. "The stem family and the developmental cycle of the peasant household: An eighteenth-century Austrian example." The American Historical Review 77(2):398.

Chayanov, A.V. 1986. The Theory of Peasant Economy. Madison: University of Wisconsin Press.

Czap, P. 1982. "The perennial multiple family household, Mishino, Russia 1782-1858." Journal of Family History 7(1):5-26.

—. 1983. "'A large family: the peasant's greatest wealth': serf households in Mishino, Russia, 1814-1858." Pp. 105-152 in Family Forms in Historic Europe, edited by R. Wall, J. Robin, and P. Laslett. Cambridge: Cambridge University Press.

Engelen, T. and A.P. Wolf. 2005. Marriage and the Family in Eurasia: Perspectives on the Hajnal Hypothesis. Amsterdam: Askant.

Erdozáin-Azpilicueta, P. and F. Mikelarena-Peña. 1998. "Labor power, social and economic differentials, and adaptive strategies of peasant households in stem-family regions of Spain." The History of the Family 3(2):155-172.

Hajnal, J. 1982. "Two kinds of preindustrial household formation systems." Population and Development Review 8(3):449-494.

Hammel, E.A. 2005. "Chayanov revisited: A model for the economics of complex kin units." Proceedings of the National Academy of Sciences 102(19):7043-7046.

Kertzer, D.I. 1984. Family Life in Central Italy, 1880-1910: Sharecropping, Wage Labor, and Coresidence. New Brunswick: Rutgers University Press.

Laslett, P. 2000. The World We Have Lost: Further Explored. London: Routledge.

Laslett, P., J. Robin, and R. Wall. 1983. Family Forms in Historic Europe. Cambridge: Cambridge University Press.

Laslett, P. and R. Wall. 1974. Household and Family in Past Time. Cambridge: Cambridge University Press.

Szoltysek, M. 2008. "Three kinds of preindustrial household formation system in historical Eastern Europe: A challenge to spatial patterns of the European family." The History of the Family 13(3):223-257.

210 Wall, R. 2001. "The transformation of the European family across the centuries." Pp. 217-241 in Family History Revisited: Comparative Perspectives, edited by R. Wall, T.K. Hareven, and J. Ehmer. Newark: University of Delaware Press.

Wood, J.W. Forthcoming. Population, Food, and Traditional Farming: A New Synthesis.

Appendix A

Valuation and Acres of Arable and Pasture

Valuation rolls are records of the taxable value of land and buildings. The Lands

Valuation (Scotland) Act of 1854 established a systematic valuation of land and buildings throughout Scotland. Prior to 1855, some valuations were recorded, but they were often sporadic and lacked consistency across locations and individual data collectors (The National Archives of

Scotland 2003). Valuation rolls were recorded annually for every burgh and county in Scotland.

Every house (including farmsteads), commercial building (stores, shops, lighthouses, etc), and plot of land (referred to as ground in the records) was recorded along with its owner (and sometimes the owner‘s local representative or factor), tenant, occupier, and value. Only the heads of households were named in these records. Annual valuation continued until 1989, after which valuation rolls only list non-domestic (commercial) properties (Cory 2004). The listing of farmsteads was ended before 1989.

Valuation rolls for the years 1855, 1861, 1871, 1881, 1891, 1901, 1911, 1921, and 1931 were collected at the Orkney County Library and Archive and entered into a computer database by the author. These years were chosen so that they would correspond with decennial census enumerations. The valuation rolls were matched to census records using farmstead names and, where possible, the name of the household head. The 1855 valuation was matched to the 1851 census enumeration, as this was the closest approximation possible and represented the earliest consistent valuation for the Northern Orkney Islands. The other valuations and census records were matched by exact year.

While valuation rolls document the taxable value of property, they may reflect the amount and quantity of land associated with a farmstead. In fact, recent work by Dr. Patricia

212 Johnson has demonstrated that the rents disputed before the Crofting Commission (established by

The Crofter‘s Holdings (Scotland) Act, 1886) are identical to the valuation of the property occupied. This analysis uses information taken from old estate (cadastral) maps, archival sources, and valuation rolls to determine the extent to which valuation is a reflection of the size of holdings in both arable and pasture and if different lairds (wealthy landlords) were translating acres into valuation using different scales.

This exploratory study uses acreages of arable and pasture listed on a historical map and in an archival document. An 1886 annotated OSGB map of Rapness estate on the southern end of the island of Westray lists the holdings of each farmstead in the margin of the map. The properties that make up this estate are listed in Table A-1. The sizes of the landholdings associated with each property were given in non-decimal form (acres, roods, falls, and ells), so these measurements were converted into decimals for use in statistical analysis (Connor and

Simpson 2004).30 Information from a different estate comes from a written document housed in the Orkney County Archives. This document, which appears to be devoted to the laird‘s accounting, dates to approximately 1891-2, and it lists the acres of arable and pasture associated with several farmsteads on the island of North Ronaldsay. These farmsteads are listed in Table

A-1. Again, land measurements were given in old units, which were translated into decimal units for the purpose of analysis.

These farmsteads were linked to their valuation records using farm names. Valuations for 1881 and 1891 were compared, and a single year, 1881, was chosen for analysis, as the

Crofting Commission revaluated many rents by 1891. This was also done to control for the effects of inflation between 1881 and 1891. The analysis described below was repeated using the

1891 valuation, and the results did not substantively change. Valuation was recorded using the old system of monetary units (pounds (£), shillings (s.), and pence (d.)). These units were

30 In Scotland, 1 acre = 4 roods, 1 rood = 40 (square) falls, and 1 fall = 6 ells.

213 decimalized for the purposes of analysis.31 From the valuation rolls, the owners of these two estates could be determined. In this period, Rapness was owned by Thomas Traill, Esq. of

Holland, Papa Westray. His estate would later go bankrupt and would be administered by a trust

(Thomson 2008). The properties on North Ronaldsay (but not the whole of the island) were owned by William Traill, Esq. of Woodwick, who resided in St. Andrews, Fife, at the time. His estate was administered locally by Benjamin Swanson of Geramount, Sanday. These landlords, although probably closely related, may not have translated acres into valuation in the same way.

To test this idea, a dummy variable for the laird is included in the analysis.32

A multiple linear regression analysis was performed on this sample of 50 properties.

Acres of arable, acres of pasture, and a dummy variable for laird were entered as predictor variables of valuation in pounds. This model was compared against a model with an added interaction term of arable and pasture. A likelihood-ratio test indicates that the addition of the

2 interaction model improves model fit (χ (1) = 6.24, p = 0.0125). Therefore, the results of the full model, which includes the interaction term, are reported in Table A-2.

Arable land has the largest significant effect on valuation. Acres of pasture, in contrast, are not a significant predictor of valuation on their own. But, the interaction of arable and pasture is significant, if small (and mostly reflects the two of three large farms in Rapness). The effect of the dummy variable for the two lairds is not significant. Overall, the fit of the model was excellent, with R2 = 0.98 (see Figure 1 for a plot of predicted versus observed valuations).

From these results, it is clear that valuation is closely related to acres of arable. The interaction of arable and pasture is also significant, although the effect is small. Lairds do not have a significant effect on valuation. This study, while preliminary, suggests that valuation rolls

31 1 pound = 20 shillings, 1 shilling = 12 pence. 32 Members of the Traill family were major landholders throughout Orkney. These two lairds are probably related, but the exact genealogical relationship cannot be inferred using the vital registration data collected by the North Orkney Population History Project. This is because these individuals were usually born and married outside of the Northern Islands.

214 are a good proxy measure for the amount and quality of land associated with a particular property or farmstead. Further, it appears to be possible to combine valuations across the holdings of multiple lairds without introducing bias, as the lairds appear to be translating acreage into valuation on the same scale. However, this sample is small relative to the size of the overall study area, and more work is needed to ensure that these trends hold for other properties and on other islands. The digitization of old maps with marked field boundaries can provide additional information about the size and quality of landholdings, as well as the use of other estate maps or archival documents that associate farmstead names with landholdings in acres.

215 Figures and Tables

Table A-1. Properities included in the analysis. All of these properties include arable land (but not necessarily pasture), with the exception of Rusk Holm, which was entirely pasture. Sheep were likely grazed on this holm, which is a general term for a small uninhabited island near a larger inhabited island.

Rapness Estate, Westray North Ronaldsay Benziecot Sholtisquoy Bu' of Rapness Garbo No. 1 Claybraes Vincoin Clifton and Mill of Rapness Westhouse Cotterochan Senness Damsay Lochend Goltiquoy South Ness Grimbist Gateside 1 Helzie Gateside 2 Hillside South Gravity Moa North Gravity Mount Pleasant Kirbist Ness Cavan Newbigging Scotsha' Perth Milldam Roadside Verracott Rusk Holm Roadside Rusland Netherbreck Sanguhar Dennishill Standcrow Bewan Sulland Senness Taftend Antabreck Tafts Tistoquoy Twinness Wasbist Whitclet Wool

216 Table A-2. Results of the multiple linear regression of valuation.

Coefficient Std. Error p-value Arable 0.6600 0.0307 <0.0001 Pasture 0.0400 0.0412 0.337 Arable X Pasture -0.0011 0.0004 0.013 Laird (0 = Rapness, 1 = North Ronaldsay) 1.0066 1.2677 0.431 Intercept -1.5068 1.1493 0.196

Figure A-1. Actual versus predicted valuation.

Actual Valuation Vs. Predicted Valuation 200 180 160 140 120 100 Rapness 80 North Predicted Valuation Predicted 60 Ronaldsay 40 20 0 0 50 100 150 200 Actual Valuation

217 References

Connor, R.D. and A.D.C. Simpson. 2004. Weights and Measures in Scotland: A European Perspective. East Linton: Tuckwell Press.

Cory, K.B. 2004. Tracing Your Scottish Ancestry. Edinburgh: Birlinn.

The National Archives of Scotland. 2003. Tracing Your Scottish Ancestors, The Official Guide: A Guide to Ancestry Research in the National Archives of Scotland. Edinburgh: Mercat Press.

Thomson, W.P.L. 2008. The New History of Orkney. Edinburgh: Birlinn Ltd.

Appendix B

Measuring Consumer-to-Producer Ratios: Do Weighting Systems Matter?

Introduction33

In preindustrial agrarian societies, the household is the primary locus of production and consumption. Household members provide the labor needed to produce enough food and other goods to meet family needs. Various measures of the size of the household labor force relative to the number of individuals supported have been proposed and employed in studies of farming societies. Generally, these measures of household composition are known as consumer-to- producer (C/P) ratios. In most cases, household members are assigned age- and sex-based weights that are intended to approximate the relative nutritional needs and energetic outputs of individuals. The production and consumption weights are then summed over all of the household members and the summed consumption weights are divided by the summed production weights to create the household-level C/P ratio.

To the author‘s knowledge, no study has systematically compared the effects of these largely arbitrary weighting systems on the performance of C/P ratios in statistical models. The purpose of this paper is to compare several measurements of C/P ratios to each other and their relative performance as predictor variables in simple models of outcomes of general interest to the study of traditional agricultural populations. Data from 19th Century Orkney, Scotland are used to construct household-level C/P ratios and outcome variables. Historical and ethnographic studies of this population provide evidence that smallholder agriculture, reliant on household

33 A version of this work was presented at the 2009 meetings of the Social Science History Association (Jennings and Wood 2009).

219 labor and with little use of wage labor, dominated production in this rural area (Jennings et al.

2009b; Murtha et al. 2008; Sparks 2007). Thus, this dataset is appropriate for this comparative study of C/P ratio measurements.

Background

Traditional farming systems are characterized by human labor inputs, the limited or lack of use of fossil fuels (for tractors, for instance), the limited use of wage laborers, and household organization of production (Netting 1993; Redfield 1989; Wolf 1966). Chayanov noted that the use of household labor distinguished smallholders from large-scale capitalist farming firms

(Chayanov 1986). In his theory of the peasant economy, the balance of workers relative to consumers was essential to understanding how families meet their needs while minimizing the drudgery of additional labor. Chayanov further noted how the relationship of household labor supply to consumption needs changes as the family develops. In his simple model, the consumer- producer (C/P) ratio is followed from the formation of a new household by a couple through their childbearing years (Table B-1). As children are born, the C/P ratio becomes sharply unfavorable, placing an economic squeeze on the household that may be mediated by increased intensity of production, decreased consumption, or the acquisition of additional labor (Figure B-1). In his consideration of the changing nature of household workers and consumers, Chayanov applied a set of age- and sex-weights to each household member, meant to represent their relative contributions to household production and nutritional needs. Other studies, expanding on

Chayanov‘s original scheme, have adopted their own sets of age- and sex- weights (Chibnik

1984; Hammel 2005; Hunt 1979; Lewis 1981). In this study, Chayanov‘s original set of weights will be used in addition to an updated set of weights proposed by Hammel in an elaboration of

Chayanov‘s model (Table B-2).

220 Regardless of weighting system chosen, several studies have found important effects of

C/P ratios. Infant mortality is sensitive to increasing C/P ratios, so that families with unfavorable economic conditions are more likely to experience the death of an infant (Campbell and Lee

1996; Sparks 2007). These findings are corroborated by evidence that children‘s anthropometric measurements, such as height, weight, and body fat, are lower for their age in families with unfavorable C/P ratios (Hagen et al. 2006). Others have found that household land use allocation changes over the course of the family life cycle, a concept analogous to Chayanovian cycles in

C/P ratios (Perz 2003). C/P ratios have been found to be predictive of the formation and dissolution of extended-family living arrangements (Jennings et al. 2009a, 2009b).

Data

The contributions of individual members to household production and actual food consumption are difficult to observe and measure in the field. However, at least one project has attempted these observations. In her study of a Yucatan Maya group that practices shifting cultivation, Kramer measured food production and consumption (Kramer 2005; Lee and Kramer

2002). While her study focused on the contribution of children to the household economy, measurements of adults were also taken. Food production was calculated by taking the mean hours per day invested by an individual in food production, transport, and preparation, weighted by efficiency (productivity/energy expenditure) relative to adults (Figure B-2). Consumption was calculated by taking the mean hours of weighted household time required to produce the food consumed by an individual per day. For use in this analysis, the consumption and production estimates are scaled so that the observed maximum is equal to 1, and a C/P ratio is calculated.

These observed contributions to production and consumption are compared to Chayanov and

Hammel‘s stylized weighting systems.

221 Individual nutritional needs and potential energetic outputs can also be inferred from laboratory measurements of aerobic capacity and published dietary guidelines. This study uses published measurements of maximum oxygen uptake, a measure of physiological work capacity, on a sample of Scandinavians as a proxy for the average work capacity of people of different ages and sexes to conduct the physically demanding work of agriculture (Figure B-3) (Rodahl 1989).

Ideal consumption requirements are calculated from published data on weight, activity level, and

FAO/WHO/UNU dietary recommendations taken from a group of rural Gambians (Table B-3)

(Ulijaszek 1995). As with Kramer‘s weights, these consumption and production estimates are scaled so that the observed maximum is equal to 1, and a ―biological‖ C/P ratio is calculated. The age and sex ―profiles‖ of each of these weighting systems are illustrated in Figures B-4 through

B-7.

The final C/P measurement used in the comparison study is the dependency ratio. This measure is frequently used as an alternative to elaborate weighting systems. To calculate the ratio, the number of dependent children and elderly adults present in the household is divided by the number of working-age adults. In this study, children 15 and under are considered dependent and adults 65 and over are considered elderly and therefore also dependent. By extension, adults ages 15-65 constitute the household labor force. Unlike the other weighting systems described above, the dependency ratio does not account for differences between the sexes or consider the potential productive contribution of children and the aged at all.

Three primary data sources are used in this study: vital registration records, individual- level census returns, and valuation rolls. Members of the North Orkney Population History

Project collected these data from the General Registrar of Scotland (GROS) and the Orkney

Library and Archive as part of an ongoing study of the population history of the northern Orkney

Islands. The decennial censuses (1851-1901) provide information on every person at home on the day of enumeration, including age, sex, and conjugal condition. From these returns, the

222 composition of the household is obtained. Information about vital events that occur in each household is taken from vital records of births and deaths, which include the name of the farmstead or house in which the individual was present. These house names, which persist over time, are used to link the census returns to the vital records. Valuation rolls, or records of the taxation or rental value of land and buildings, also list house names, which are again used to link these data to information from the censuses and vital registers. The data used in this study represent all households on the islands of Westray, Eday, Papa Westray, and Faray for which such record linkage was possible.

C/P ratios for each household are calculated from the age and sex of household members as listed in the census returns. The number of births and deaths that occur in the household for each census interval are tabulated for use as outcome variables in models designed to compare the different C/P ratios. The value of land to which each household has access is taken from the valuation rolls and used in comparative models (See Appendix A for a detailed discussion of valuation rolls). Finally, the presence of life-cycle servants, who are noted as such in census returns, and the type of family form present in the household (1=simple 0=extended or multiple, using the Hammel-Laslett classification (Hammel and Laslett 1974)) are taken from census records and used in comparative models.

Methods

Five C/P ratios, Chayanov, Hammel, Kramer, biological, and dependency ratio, were calculated using data from Northern Orkney during a period when traditional agriculture dominated the local economy (1851-1901). These measures are highly correlated with each other, with correlation coefficients ranging from 0.61 to 0.89 (Table B-4). Scatter plots of the

C/P measurements provide some evidence that they are linearly related (Figure B-8).

223 The C/P ratios, after scaling so that the maximum weight=1 and standardizing, are used as predictor variables in simple models. Controls for household size are included (except for the model of household size as an outcome). This is important when using ratio measurements, because the relative impact of the loss or addition of a household member on the C/P ratio is influenced by the size of the household. The simple models have one outcome variable and C/P ratio and household size as predictor variables. The models can be categorized into demographic outcomes (births and deaths), household level outcomes (household size, family form), and economic outcomes (land value, presence of servant). The coefficients of each C/P measurement are compared.

Results and Discussion

Figure 9 summarizes the coefficients and p-values of the demographic models.

Generally, the fit of these models is poor, but this is unsurprising given the complicated mechanisms that influence household demographic outcomes. In these models, the dependency ratio and the Kramer weights appear to differ the most from the other C/P ratios. Figure B-10 summarizes the coefficients and p-values of the household level outcomes. In these models, the fit is better, and the C/P ratios are more consistent in their effects on the estimation of coefficients. Figure B-11 summarizes the coefficients and p-values of the models of economic outcomes. In these models, the C/P ratios are consistent in direction of effect and similar in the size of the effect. As a rough estimate of the congruence of these measurements, absolute difference of the coefficient of each C/P ratio from the mean value was calculated for each model.

These differences were summed to provide a single value to compare the performance of each

C/P measurement relative to the mean. The results are presented in Table B-5. The dependency ratio differs the most from the mean, which is not surprising given that it is the simplest measure

224 of C/P ratio and makes the fewest assumptions about age and sex weights. The dependency ratio also produces a coefficient that is of a different sign than the majority of other C/P ratios in one case. The biological weights differ the least from the mean. In addition, it is the only C/P ratio to produce coefficient estimates with direction of effects consistent with the majority of the C/P ratios in all models.

Despite these observed differences in performance of various C/P ratios as predictor variables, overall the measurements are remarkably consistent. Indeed, even the C/P measurement with the crudest assessment of individual contributions to the household economy, the dependency ratio, performs relatively well. While researchers should always explicitly state their particular choice of weights when constructing C/P ratios, it is unlikely that the specific choice of weights should drastically affect model results, particularly when other important control variables are added to the models.

This analysis represents a preliminary assessment of the effects of different C/P ratio measurements on the performance of the variable in statistical models. While a systematic sensitivity analysis will produce clearer estimates of these effects, this study is a first step in comparing commonly used weighting systems with observational measurements and biological approximations. The results presented here suggest that C/P ratios are related more to population structure than to the specific choice of age- and sex- based weights, given that they roughly approximate what scientists know about individual levels of production and consumption across the life course. Simulation studies that allow for varying population structures offer a test of this hypothesis and remain a topic for future analysis.

225 Tables and Figures

Table B-1. Chayanov‘s model of change in household composition (Chayanov 1986, p. 58).

Figure B-1. Chayanov‘s model of household cycles in C/P ratio (Chayanov 1986, p. 59).

226 Table B-2. Hammel‘s modification of Chayanov‘s weights used to calculate C/P ratios (after Hammel 2005). Production Consumption Male Female Male Female Age Units Age Units Age Units Age Units 0-5 0 0-5 0 0-2 0.1 0-2 0.1 5-7 0.1 5-6 0.2 2-5 0.3 2-5 0.3 7-9 0.2 6-10 0.5 5-9 0.5 5-6 0.5 9-12 0.5 10-15 0.7 9-12 0.7 6-10 0.7 12-15 0.9 15-20 0.7 12-15 0.8 10-12 0.8 15-50 1 20-60 0.8 15-50 1 12-60 0.8 50+ 0.8 60+ 0.7 50+ 0.8 60+ 0.7

Figure B-2. Observational study of weighted contribution to production and consumption by age and sex (Kramer 2005, p. 128).

227 Figure B-3. Physiological work capacity by age and sex (Rodahl 1989, p. 20).

Table B-3. Energy requirements (MJ/day) by age and sex of rural Gambians (Ulijaszek 1995, p. 160). Males Females

Energy Requirement Age (MJ/day) 1 3.7 3.5 2 4.9 4.2 3 5.9 5.3 4 6.1 5.6 5 6.3 5.8 6 6.5 6.0 7 6.8 6.3 8 7.2 6.7 9 7.7 7.2 10 8.0 7.7 11 8.4 8.0 12 8.9 8.4 13 9.1 8.8 14 9.8 9.1 15 10.4 9.4 16 10.8 9.6 17 11.3 9.6 18-30 12.2 9.0 30-60 11.9 9.0

228 Figure B-4. Profiles of male production weights by age.

Figure B-5. Profiles of female production weights by age.

229 Figure B-6. Profiles of male consumption weights by age.

Figure B-7. Profiles of female consumption weights by age.

230

Table B-4. Correlation coefficients for C/P ratios. All correlations are statistically significant.

Dependency Hammel Chayanov Biological Kramer Ratio Hammel 1 Chayanov 0.61 1 Biological 0.72 0.89 1 Kramer 0.73 0.78 0.88 1 Dependency Ratio 0.63 0.76 0.74 0.69 1

N=2575

231

Figure B-8. Scatter plots of C/P measurements.

6 6

4 4

2 2

0 0

Standardized values of (ham) Standardized values of (ham)

-2 -2 0 5 10 15 -2 0 2 4 6 8

Standardized values of (chay) Standardized values of (biostd)

6

15

4

10

2

5

0

Standardized values of (ham)

Standardized values of (chay)

0 -2 0 5 10 -2 0 2 4 6 8

Standardized values of (dr65) Standardized values of (biostd)

6

15

4

10

2

5

0

Standardized values of (ham)

Standardized values of (chay)

0 -2 -2 0 2 4 6 0 5 10

Standardized values of (kramstd) Standardized values of (dr65)

15

8

6

10

4

2

5

Standardized values of (chay)

0

Standardized values of (biostd)

0 -2 -2 0 2 4 6 0 5 10

Standardized values of (kramstd) Standardized values of (dr65)

8

6

6

4

4

2

2

0

0

Standardized values of (biostd)

Standardized values of (kramstd)

-2 -2

-2 0 2 4 6 -2 0 2 4 6 8 Standardized values of (kramstd) Standardized values of (biostd)

232 Figure B-9. Coefficients of models of demographic outcomes (blue=statistically significant, green=not statistically significant). Coefficients in models of births Coefficients in models of deaths 0.15 under age 15 0.016 0.1 0.014 0.012 0.05 0.01 0.008 0 0.006 0.004 Coefficent 0.002

-0.05 0 Coefficent -0.002 -0.1 -0.004

-0.15

-0.2

Coefficients in models of deaths Coefficients in model deaths over age 65 ages 15-65 0.04 0 0.02 -0.005 0 -0.01 -0.02

-0.04 -0.015 Coefficent -0.06 Coefficent -0.02 -0.08 -0.025 -0.1 -0.12 -0.03

Coefficients in models of all deaths 0

-0.02

-0.04

-0.06

Coefficent -0.08

-0.1

-0.12

233 Figure B-10. Coefficients of models of household-level outcomes (blue=statistically significant, green=not statistically significant). Coefficients in models of Coefficients in models of family forms (simple family=1) household size 1.2 1.6 1 1.4 1.2 0.8 1 0.6 0.8

0.6 Coefficent Coefficent 0.4 0.4 0.2 0.2 0 0

Figure B-11. Coefficients of models of economic outcomes (blue=statistically significant, green=not statistically significant). Coefficients in models of Coefficients in models of land presence of servants (present=1) value 0 0

-0.1 -0.5

-0.2 -1 -0.3

Coefficent -1.5 Coefficent -0.4

-2 -0.5

-0.6 -2.5

Table B-5. Total absolute differences of C/P measurement from mean coefficients for all simple models. Hammel 0.48 Chayanov 0.74 Biological 0.44 Kramer 0.91 Dependency Ratio 1.24

234 References

Campbell, C. and J.Z. Lee. 1996. "A death in the family: Household structure and mortality in rural Liaoning: Life-event and time-series analysis, 1792–1867." The History of the Family 1(3):297-328.

Chayanov, A.V. 1986. The Theory of Peasant Economy. Madison: University of Wisconsin Press.

Chibnik, M. 1984. "A cross-cultural examination of Chayanov's theory." Current Anthropology 25(3):335.

Hagen, E.H., H.C. Barrett, and M.E. Price. 2006. "Do human parents face a quantity-quality tradeoff? Evidence from a Shuar community." American Journal of Physical Anthropology 130(3):405-418.

Hammel, E.A. 2005. "Chayanov revisited: A model for the economics of complex kin units." Proceedings of the National Academy of Sciences 102(19):7043-7046.

Hammel, E.A. and P. Laslett. 1974. "Comparing household structure over time and between cultures." Comparative Studies in Society and History 16(1):73-109.

Hunt, D. 1979. "Chayanov's model of peasant household resource allocation." Journal of Peasant Studies 6(3):247-285.

Jennings, J.A., C.S. Sparks, J.W. Wood, P.L. Johnson, T.M. Murtha, and S.A. Matthews. 2009a. "Dissolution of extended-family households in Northern Orkney, Scotland, 1851-1901." Presented at the annual meeting of the Population Association of America, Detroit, MI.

—. 2009b. "Formation and dissolution of extended-family households in Orkney, Scotland, 1851- 1901." Presented at the meeting of the International Union for the Scientific Study of Population, Marrakech, Morocco.

Jennings, J.A. and J.W. Wood. 2009. "Measuring Consumer-to-Producer Ratios: Do Weighting Systems Matter?" Presented at the annual meeting of the Social Science History Association, Long Beach, CA.

Kramer, K.L. 2005. Maya Children: Helpers At The Farm. Cambridge: Harvard University Press.

Lee, R.D. and K.L. Kramer. 2002. "Children's economic roles in the Maya family life cycle: Cain, Caldwell, and Chayanov revisited." Population and Development Review 28(3):475-499.

Lewis, J.V.D. 1981. "Domestic labor intensity and the incorporation of Malian peasant farmers into localized descent groups." American Ethnologist 8(1):53-73.

Murtha, T.M., P.L. Johnson, J.W. Wood, S.A. Matthews, C.S. Sparks, and J.A. Jennings. 2008. "Historical demography, oral history, settlement archaeology, and landscape ecology: The North Orkney Population History Project." Presented at the annual meeting of the European Population Conference, Barcelona, Spain.

235 Netting, R.M. 1993. Smallholders, Householders: Farm Families and the Ecology of Intensive, Sustainable Agriculture. Stanford: Stanford University Press.

Perz, S.G. 2003. "Household demography and land use allocation among small farms in the Brazilian Amazon." Human Ecology Review 9(2):1-16.

Redfield, R. 1989. The Little Community and Peasant Society and Culture. Chicago, IL: University of Chicago Press.

Rodahl, K. 1989. The Physiology of Work. Philadelphia: Taylor and Francis.

Sparks, C.S. 2007. Households, Land, and Labor: Population Dynamics in the Northern Orkney Islands, Scotland, 1851 to 2003. Ph.D. Dissertation, Anthropology and Population Research Institute, Pennsylvania State University.

Ulijaszek, S.J. 1995. Human Energetics in Biological Anthropology. Cambridge: Cambridge University Press.

Wolf, E.R. 1966. Peasants. Englewood Cliffs: Prentice-Hall.

Appendix C

A Contemporary Description of Servants and Hiring Fairs

This account of servants is given by a parson living in the lake country of England around

1890 (Watson 1894). It speaks to many of the characteristic features of the kind of servants under study. He writes (pp. 94-100):

In describing the lives and homes of the yeomen,34 I remarked that their sons and daughters went out to service. This spirit of honest independence contributed much to their success, and was one of their chief characteristics. And because it was so I am constrained to describe more at length the virtues of these hard- working sons and daughters, and to hold them up (at least in agricultural matters) for imitation and respect. The lives of this class, except by being bettered, have changed less than those of their superiors, and I shall endeavor to describe them as they exist pretty much to-day, not as they existed in times gone by. All our bread, all our wealth, comes from the land, and sometimes I am afraid we are apt to under-rate those who by their labour win it from the soil.

The farm labourer of the dales, then (and he is more often than not the son of a small farmer or yeoman), is nothing akin to his southern brother. And it is probably no exaggeration to say that he is superior to him in every way in which comparison is possible. The southerner seems unable to lift himself above his surroundings, whilst the northerner almost invariable succeeds. Much of this is probably owing to the fact that he is early sent to school, but at fourteen leaves home to earn his own living. He has been well schooled, in a way, and looks forward to ―service.‖ At the half-yearly hiring—Whitsuntide or Martinmas— after he has attained his ―first majority,‖ he goes to the nearest country town and stands in the market-place. He is attired in a brand new suit, with a capacious necktie of green and red. These articles he has donned upon the memorable morning, and as a gift from his parents they constitute his start in life. The country barber has left his head pretty much as the modern reaper leaves the stubble, and has not stinted him of grease for his money. As an outward and visible sign of his intention, the lad sticks a straw in his mouth and awaits the issue. For the first hour or so he keeps his eyes bent to the pavement, as though to read the riddle of his life there, but presently gains confidence to look about him. After waiting a greater part of the morning and seeing many of his fellow- men and maid-servants hired, he is accosted by a stalwart yeoman, who inquires if he wants a ―spot‖—a place, a situation. The lad replies that he does; that he is

34 Yeomen are small freeholders (my comment).

237 willing to do anything; and the he will engage for £4 the half-year—―if it pleases.‖ A bargain is soon struck, and the stalwart urchin from the ―fell-heads‖ marches off to lose himself in the giddy gaiety of the Fair. If ultimately he likes his ―place,‖ and is well and kindly treated, we may not see him again for a couple of years. During this time he has made himself generally useful, has become a good milker, and has shone conspicuously at hay and harvest. He has proved himself a ―fine lad,‖ and has had his wages raised by way of reward. At sixteen or seventeen he is stalwart enough to hire as a man, and now his wages are doubled; he asks and obtains £12 for the year, or even £14 if entering upon the summer half. The farm servants of the dales ―live in,‖ and have all found [sic]. They are well fed, well housed, and have their meals at the master‘s table. But if well fed they are hard worked, and in summer they often rise as early as three or four in the morning. In these parts, which constitute a vast grazing district, the labour of the farm servant is much more general and interesting than that of his southern brother, where the land is arable….A few years ago, during the flourishing times of agriculture, the northern labourer obtained from £35 to £40 per annum, still, of course, ―living in‖; a few picked men could even command £45. Now, however, that we are come upon times of depression, the best men are glad to work for the first named sum. In proportion, the girls are much better off in the matter of wages than the men. There is probably less competition among them, owing to the fact that there is a great temptation for country girls to migrate and enter service in provincial towns. Here they are not so hard worked as in the farm-houses, and have the satisfaction of being engaged in what they esteem a much more ―genteel‖ occupation.

Many of the men, when about thirty years of age, are able to take small farms of their own. Nearly all the statesmen‘s sons do this, and probably without any outside help; for, as a class, these labourers are not only industrious but thrifty. I knew a man who had saved £120, which sum he had divided and deposited into three banks. This was his whole wealth, and he told me he did not want to lose his hard-earned savings if the banks should ―break.‖ His object was to acquire a small farm, and he has now succeeded.

From the fact of ―living in,‖ as nearly all the valley servants do, it need hardly be said that early marriages are rare. All the better men look forward to the time when they can have a farm of their own; and when they obtain a holding, they then look for a wife. This fact alone speaks well for their thrift; but it has its dark side. How far the two things are connected may be a matter of speculation; but it is notorious that the number of illegitimate children in the north is far above average, and most of them undoubtedly are born of the agricultural classes. The registers of the country churches abundantly prove this. Still, it is pleasing to be able to record the fact that in the dales, sooner or later, those who have been wronged ―are made honest women of‖ by marriage.

Appendix D

An Example Life History of a Farm Servant

Unfortunately, the data available from the North Orkney Population History Project do not allow for the detailed reconstruction of individual life courses with respect to employment and migration. To provide an example of the life history of a life-cycle servant, an autobiographical account is more appropriate. Fred Kitchen was a life-cycle servant in

Yorkshire, England. The major events of his life, reconstructed (by me) from his account,

Brother to the Ox, are summarized here (Kitchen 1981). He went on to write several more books on country life, poems, and a novel. After becoming a journalist and radio broadcaster, Fred died in 1969.

1891—Born in Nottinghamshire. Soon after, his family moves to South Yorkshire, England. Father works as a cattleman on a large estate and received wages and a free cottage. 1902—Father dies when Fred was age 11. Because of his father‘s death, Fred cannot be apprenticed to his uncle, a joiner, as originally planned. 1903—Fred‘s older sister leaves school at age 13 and becomes a domestic servant. The family must vacate the cottage on the estate, so they move to a house in the nearby village. Fred‘s mother makes a living doing needlework and serving tea to travelers. 1904—Fred leaves school at age 13 and works as a ―day lad‖ on a 130 acre farm. Day lads work days, but live at home. 1904—A servant leaves the farm at Martlemas [Martinmas], so Fred decides to take his place and ―live in.‖ 1906—Fred leaves the farm to work on the railroad at age 15. 1907—Fred rejoins agricultural service at a different farm at the next Martlemas hiring fair. He is now a horse lad and ploughman. 1908—At age 17, Fred is now a wagoner (horseman). This is a more respected/skilled position. 1909—Fred begin to court the village doctor‘s nursemaid, Helen. Also a life-cycle servant, Helen‘s father is a coachman. 1910—The farmer‘s wife dies, and the farmer decides to retire and live with his married daughter. Fred moves back home, and pays his mother for room and board while he works as a carter. Returns to farm service and works as a horseman. 1912—Leaves farm service to work at the colliery. Later promoted to sulphate-house attendant. 1914—Does not have to serve during WWI because of the importance of coal-workers to the war effort. 1915—Marries Helen, after 5-year courtship. They took a house in the village, using Fred‘s savings. 1916—First daughter, Margaret, is born. 1918—First son, William, is born.

239 1920—Helen dies of the flu. Fred gives up lease of the smallholding and moves back in with his mother. 1921—Coal strike wipes out Fred‘s savings, so he could not buy a smallholding of his own. 1925—Fred marries a second wife, Lizzie, and returns to farm labor. He and his family live in a cottage attached to the farm. This farm has a dairy, and Fred becomes the ―milk- roundsman‖ and makes deliveries in Sheffield. 1927—Youngest child born. 1927—Switches farms to obtain a better cottage, as the old cottage was unhealthy for a newborn. 1936—The farm is sold, and Fred stayed on with the new farmer and goes back to being a wagoner. 1939—Writes his autobiography, Brother to the Ox, encouraged by classes offered by the Workers‘ Education Association.

240 References

Alderson, A.S. and S.K. Sanderson. 1991. "Historic European household structures and the capitalist world-economy." Journal of Family History 16(4):419-432.

Alter, G. 1988. Family and the Female Life Course: The Women of Verviers, Belgium, 1849- 1880. Madison: University of Wisconsin Press.

Anderson, M. 1971. Family Structure in Nineteenth Century Lancashire. Cambridge: Cambridge University Press.

—. 1995. Approaches to the History of the Western Family, 1500-1914. Cambridge: Cambridge University Press.

Anderson, M. and D.J. Morse. 1993a. "High fertility, high emigration, low nuptiality: Adjustment processes in Scotland's demographic experience, 1861–1914, Part I." Population Studies 47(1):5-25.

—. 1993b. "High fertility, high emigration, low nuptiality: Adjustment processes in Scotland's demographic experience, 1861–1914, Part II." Population Studies 47(2):319-343.

Andorka, R. and S. Balazs-Kovacs. 1986. "The social demography of Hungarian villages in the eighteenth and nineteenth centuries (with special attention to Sarpilis, 1792-1804)." Journal of Family History 11(2):169-192.

Arru, A. 1990. "The distinguishing features of domestic service in Italy." Journal of Family History 15(4):547-566.

Ashmore, W. and R.R. Wilk. 1988. ―Household and community in the Mesoamerican past.‖ Pp. 1-27 in Household and Community in the Mesoamerican Past, edited by W. Ashmore and R.R. Wilk. Albuquerque: University of New Mexico Press.

Bengtsson, T., C. Campbell, and J.Z. Lee. 2004. Life Under Pressure: Mortality and Living Standards in Europe and Asia, 1700-1900. Cambridge: MIT.

Bengtsson, T. and G.P. Mineau. 2008. Kinship and Demographic Behavior in the Past. New York: Springer.

Bengtsson, T. and O. Saito. 2003. Population and Economy: From Hunger to Modern Economic Growth. Oxford: Oxford University Press.

Berkner, L.K. 1972. "The stem family and the developmental cycle of the peasant household: An eighteenth-century Austrian example." The American Historical Review 77(2):398.

—. 1975. "The use and misuse of census data for the historical analysis of family structure." Journal of Interdisciplinary History 5(4):721-738.

241 Blackstone, W. 1979. Commentaries on the Laws of England: A Facsimile of the First Edition of 1765-1769. Chicago: University of Chicago Press.

Bonneuil, N. 1997. Transformation of the French Demographic Landscape, 1806-1906. Oxford: Clarendon.

Boulton, J. 2000. "'It is extreme necessity that makes me do this': Some "survival strategies" of pauper households in London's West End during the early eighteenth century." Pp. 47-70 in Household Strategies for Survival, 1600-2000: Fission, Faction, and Cooperation, edited by L. Fontaine and J. Schlumbohm. Cambridge: Cambridge University Press.

Brennan, E.R. 1979. Kinship, Demographic, Social and Geographic Characteristics of Mate Choice in a Small Human Population. Dissertation, Anthropology, Pennsylvania State University.

Buchan, D. 1984. "The expressive culture of nineteenth-century Scottish farm servants." Pp. 226- 242 in Farm Servants and Labour in Lowland Scotland, 1770-1914, edited by T.M. Devine. Edinburgh: John Donald.

Bumpass L.L. and R.K. Raley. 1995. ―Redefining single-parent families: Cohabitation and changing family reality.‖ Demography 32(1):97-109.

Burch, T.K. 1970. "Some demographic determinants of average household size: An analytic approach." Demography 7(1):61-69.

Campbell, C. and J.Z. Lee. 1996. "A death in the family: Household structure and mortality in rural Liaoning: Life-event and time-series analysis, 1792–1867." The History of the Family 1(3):297-328.

Carter, A.T. 1984. "Household histories." Pp. 44-83 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

Chayanov, A.V. 1931. ―The socio-economic nature of peasant farm economy.‖ Pp. 144-147 in A Systematic Source Book in Rural Sociology, Volume II, edited by P.A. Sorokin, C.C. Zimmerman, and C.J. Galpin. Minneapolis: University of Minnesota Press.

—. 1986. The Theory of Peasant Economy. Madison: University of Wisconsin Press.

Chibnik, M. 1984. "A cross-cultural examination of Chayanov's theory." Current Anthropology 25(3):335.

Coale, A.J. 1965. "Appendix: Estimates of average size of household." in Aspects of the Analysis of Family Structure, edited by M. Levy and A.J. Coale. Princeton: Princeton University Press.

Connor, R.D. and A.D.C. Simpson. 2004. Weights and Measures in Scotland: A European Perspective. East Linton: Tuckwell Press.

242 Cooper, S.M.I. 2004. "From family member to employee: Aspects of continuity and discontinuity in English domestic service, 1600-2000." Pp. 277-296 in Domestic Service and the Foundation of European Identity, edited by A. Fauve-Chamoux. New York: Peter Lang.

—. 2005. "Service to servitude? The decline and demise of life-cycle service in England." The History of the Family 10(4):367-386.

Cory, K.B. 2004. Tracing Your Scottish Ancestry. Edinburgh: Birlinn.

Czap, P. 1982. "The perennial multiple family household, Mishino, Russia 1782-1858." Journal of Family History 7(1):5-26.

—. 1983. "'A large family: the peasant's greatest wealth': serf households in Mishino, Russia, 1814-1858." Pp. 105-152 in Family Forms in Historic Europe, edited by R. Wall, J. Robin, and P. Laslett. Cambridge: Cambridge University Press.

Das Gupta, M. 1997a. "Kinship systems and demographic regimes." Pp. 36-52 in Anthropological Demography: Toward a New Synthesis, edited by D.I. Kertzer and T. Fricke. Chicago: University of Chicago Press.

—. 1997b. "Socio-economic status and clustering of child deaths in rural Punjab." Population Studies 51(2):191-202.

—. 1999. "Lifeboat versus corporate ethic: Social and demographic implications of stem and joint families." Social Science and Medicine 49(2):173-184.

Devine, T.M. 1984a. "Introduction: Scottish farm service in the Agricultural Revolution." Pp. 1-9 in Farm Servants and Labour in Lowland Scotland, 1770-1914, edited by T.M. Devine. Edinburgh: John Donald.

—. 1984b. "Women workers, 1850-1914." Pp. 98-123 in Farm Servants and Labour in Lowland Scotland, 1770-1914, edited by T.M. Devine. Edinburgh: John Donald.

Dodgshon, R.A. 1994. "Budgeting for survival: Nutrient flow and traditional Highland farming." Pp. 83-93 in The History of Soils and Field Systems, edited by S. Foster and T.C. Smout. Aberdeen: Scottish Cultural Press.

Dribe, M. 2000. Leaving Home in a Peasant Society: Economic Fluctuations, Household Dynamics and Youth Migration in Southern Sweden, 1829 - 1866. Sodertalje: Almqvist & Wiksell International.

Dunbar, J.G. 1966. The Historic Architecture of Scotland. London: Batsford.

Elder, G.H. 1978. "Family history and the life course." Pp. 17-64 in Transitions: The Family and the Life Course in Historical Perspective, edited by T.K. Hareven. New York: Academic Press.

Engelen, T. and A.P. Wolf. 2005a. "Introduction: Marriage and the family in Eurasia. Perspectives on the Hajnal hypothesis." Pp. 15-36 in Marriage and the Family in

243 Eurasia: Perspectives on the Hajnal Hypothesis, edited by T. Engelen and A.P. Wolf. Amsterdam: Aksant.

—. 2005b. Marriage and the Family in Eurasia: Perspectives on the Hajnal Hypothesis. Amsterdam: Askant.

Erdozáin-Azpilicueta, P. and F. Mikelarena-Peña. 1998. "Labor power, social and economic differentials, and adaptive strategies of peasant households in stem-family regions of Spain." The History of the Family 3(2):155-172.

Ewan, E. 2004. "Mistresses of themselves? Female domestic servants and by-employments in sixteenth century Scottish towns." Pp. 411-434 in Domestic Service and the Formation of European Identity, edited by A. Fauve-Chamoux. New York: Peter Lang.

Fauve-Chamoux, A. and R. Wall. 2005. "Domestic servants in comparative perspective: Introduction." The History of the Family 10:345-354.

Fenton, A. 1976. Scottish Country Life. Edinburgh: John Donald.

—. 1997. The Northern Isles: Orkney and Shetland. East Linton: Tuckwell Press.

Fenton, A. and B. Walker. 1981. The Rural Architecture of Scotland. Edinburgh: John Donald.

Firth, R. 1966. Housekeeping among Malay Peasants. 2nd Edition. London: Athlone Press.

Fjellman, S.M. 1977. "The Akamba domestic cycle as Markovian process." American Ethnologist 4(4):699-713.

Flandrin, J.L. 1976. Families in Former Times: Kinship, Household and Sexuality. Cambridge: Cambridge University Press.

Flinn, M.W., J. Gillespie, N. Hill, A. Maxwell, R. Mitchison, and C. Smout. 1977. Scottish Population History: From the 17th Century to the 1930s. Cambridge: Cambridge University Press.

Fontaine, L. and J. Schlumbohm. 2000. "Household strategies for survival: An introduction." Pp. 1-18 in Household Strategies for Survival, 1600-2000: Fission, Faction, and Cooperation, edited by L. Fontaine and J. Schlumbohm. Cambridge: Cambridge University Press.

Fortes, M. 1958. "Introduction." Pp. 1-14 in The Developmental Cycle in Domestic Groups, edited by J. Goody. Cambridge: Cambridge University Press.

Gaunt, D. 1978. "Household typology: Problems--methods--results." Pp. 69-83 in Change and Change: Social and Economic Studies in Historical Demography in the Baltic Area, edited by S. Akerman, H.C. Johansen, and D. Gaunt. Odense: Odense University Press.

Goldberg, P.J.P. 1992. Women, Work, and Life Cycle in a Medieval Economy: Women in York and Yorkshire c. 1300-1520. New York: Oxford University Press, USA.

244 Goody, J. 1958. The Developmental Cycle in Domestic Groups. Cambridge: Cambridge University Press.

—. 1972. "The evolution of the family." Pp. 103-124 in Household and Family in Past Time, edited by P. Laslett and R. Wall. Cambridge: Cambridge University Press.

—. 1976. Production and Reproduction: A Comparative Study of the Domestic Domain. Cambridge: Cambridge University Press.

—. 1996. "Comparing family systems in Europe and Asia: Are there different sets of rules?" Population and Development Review 22(1):1-20.

Goubert, P. and G. Denault. 1977. "Family and province: A contribution to the knowledge of family structures in early modern France." Journal of Family History 2(3):179-195.

Gray, J. 2006. "Gender composition and household labour strategies in pre-Famine Ireland." The History of the Family 11(1):1-18.

Gray, M. 1957. The Highland Economy, 1750-1850. Edinburgh: Oliver.

—. 1984. "Farm workers in North-east Scotland." Pp. 10-28 in Farm Servants and Labour in Lowland Scotland, edited by T.M. Devine. Edinburgh: John Donald.

Gray, R.F. 1964. ―Introduction.‖ Pp. 1-33 in The Family Estate in Africa: Studies in the Role of Property in Family Structure and Lineage Continuity, edited by R.F Gray and P.H. Gulliver. London: Routledge and Kegan Paul.

Hagen, E.H., H.C. Barrett, and M.E. Price. 2006. "Do human parents face a quantity-quality tradeoff?: Evidence from a Shuar community." American Journal of Physical Anthropology 130(3):405-418.

Hajnal, J. 1953. "Age at marriage and proportions marrying." Population Studies 7(2):111-136.

—. 1965. "European marriage patterns in perspective." Pp. 101-143 in Population in History, edited by D.V. Glass and D.E.C. Eversley. London: Arnold.

—. 1982. "Two kinds of preindustrial household formation systems." Population and Development Review 8(3):449-494.

Hammel, E.A. 1972. "The zadruga as process." Pp. 335-373 in Household and Family in Past Time, edited by P. Laslett and R. Wall. Cambridge: Cambridge University Press.

—. 1984. "On the *** of studying household form and function." Pp. 29-43 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

—. 2005. "Chayanov revisited: A model for the economics of complex kin units." Proceedings of the National Academy of Sciences 102(19):7043-7046.

245 Hammel, E.A. and P. Laslett. 1974. "Comparing household structure over time and between cultures." Comparative Studies in Society and History 16(1):73-109.

Harrison, W. 1968. The Description of England. Ithaca: Cornell University Press.

Hasbach, W. 1908. A History of the English Agricultural Labourer. London: PS King & Son.

Hirth, K.G. 1993. ―The household as an analytical unit: Problems in method and theory.‖ Pp. 21- 36 in Prehispanic Domestic Units in Western Mesoamerica: Studies of the Household, Compound, and Residence, edited by R.S. Santley and K.G. Hirth. Boca Raton: CRC Press.

—. 2009. ―Craft production, household diversification, and domestic economy in Prehispanic Mesoamerica.‖ Archaeological Papers of the American Anthropological Association 19(1):13-32.

Howatson, W. 1984. "Grain Harvesting and Harvesters." Pp. 124-142 in Farm Servants and Labour in Lowland Scotland, 1770-1914, edited by T.M. Devine. Edinburgh: John Donald.

Hunt, D. 1979. "Chayanov's model of peasant household resource allocation." Journal of Peasant Studies 6(3):247-285.

Itzigsohn, J. 1995. ―Migrant remittances, labor markets, and household strategies: A comparative analysis of low-income household strategies in the Caribbean Basin.‖ Social Forces 74(2):633-655.

Janssens, A. 2002. Family and Social Change: The Household as a Process in an Industrializing Community. Cambridge: Cambridge University Press.

Jennings, J.A. 2010. "Household level predictors of the presence of life-cycle servants in northern Orkney, Scotland, 1851-1901." Presented at the annual meeting of the Population Association of America, Dallas, TX.

Jennings, J.A., C.S. Sparks, J.W. Wood, P.L. Johnson, T.M. Murtha, and S.A. Matthews. 2009a. "Dissolution of extended-family households in Northern Orkney, Scotland, 1851-1901." Presented at the annual meeting of the Population Association of America, Detroit, MI.

—. 2009b. "Formation and dissolution of extended-family households in Orkney, Scotland, 1851- 1901." Presented at the meeting of the International Union for the Scientific Study of Population, Marrakech, Morocco.

Jennings, J.A., and Wood, J.W. 2008. ―Multi-census methods for anthropological demography: Applications to the Northern Islands of Orkney, Scotland. Presented at the annual meeting of the Human Biology Association, Columbus, OH.

—. 2009. "Measuring Consumer-to-Producer Ratios: Do Weighting Systems Matter?" Presented at the annual meeting of the Social Science History Association, Long Beach, CA.

246 Jonker, M.A. and A.W. van der Vaart. 2007. "Correcting missing-data bias in historical demography." Population Studies 61(1):99-113.

Kahk, J., H. Palli, and H. Uibu. 1982. "Peasant family and household in Estonia in the eighteenth and the first half of the nineteenth centuries." Journal of Family History 7(1):76-88.

Kasakoff, A.B. and J.W. Adams. 1995. "The effect of migration on ages at vital events: A critique of family reconstitution in historical demography." European Journal of Population 11(3):199-242.

Katz, M.B. 1975. The People of Hamilton, Canada West: Family and Class in a Mid-nineteenth- Century City. Cambridge: Harvard University Press

Kertzer, D.I. 1984. Family Life in Central Italy, 1880-1910: Sharecropping, Wage Labor, and Coresidence. New Brunswick: Rutgers University Press.

—. 1989. "The joint family household revisited: Demographic constraints and household complexity in the European past." Journal of Family History 14(1):1-15.

Kertzer, D.I. and M. Barbagli. 2002a. The History of the European Family: Family life in the long nineteenth century (1789-1913). New Haven: Yale University Press.

—. 2002b. The History of the European Family: Family life in the long nineteenth century, 1789- 1913. New Haven: Yale University Press.

Kertzer, D.I. and D.P. Hogan. 1990. "Household organization and migration in nineteenth- century Italy." Social Science History 14(4):483-505.

Kitchen, F. 1981. Brother to the Ox: The Autobiography of a Farm Labourer. Horsham, Sussex: Caliban Books.

Kramer, K.L. 2005. Maya Children: Helpers At The Farm. Cambridge: Harvard University Press.

Kunstadter, P. 1984. "Cultural ideals, socioeconomic change, and household composition: Karen, Lua', Hmong, and Thai in Northwestern Thailand." Pp. 299-329 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

Kussmaul, A. 1981a. Servants in Husbandry in Early Modern England. Cambridge: Cambridge University Press.

Kussmaul, A.S. 1981b. "The ambiguous mobility of farm servants." The Economic History Review 34(2):222-235.

Landsteiner, E. 1999. "Household, family, and economy among wine-growing peasants The case of lower Austria in the first half of the nineteenth century." The History of the Family 4(2):113-135.

Laslett, P. 1969. ―The household in England over three centuries.‖ Population Studies. 23(2):199-223.

247

—.1977a. Family life and illicit love in earlier generations: Essays in historical sociology. New York: Cambridge University Press.

—. 1977b. "Introduction." Pp. 23-89 in Household and Family in Past Time, edited by P. Laslett and R. Wall. Cambridge: Cambridge Univ. Press.

—. 1983. "Family and household as work group and kin group: areas of traditional Europe compared." Pp. 513-564 in Family Forms in Historic Europe, edited by R. Wall, J. Robin, and P. Laslett. Cambridge: Cambridge University Press.

—. 1984a. "The family as a knot of individual interests." Pp. 353-382 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

—. 1984b. The World We Have Lost New York: Scribner.

—. 1988. "Family, kinship and collectivity as systems of support in pre-industrial Europe: A consideration of the ‗nuclear-hardship‘ hypothesis." Continuity and Change 3(2):153- 175.

—. 2000. The World We Have Lost: Further Explored. London: Routledge.

Laslett, P., J. Robin, and R. Wall. 1983. Family Forms in Historic Europe. Cambridge: Cambridge University Press.

Laslett, P. and R. Wall. 1974. Household and Family in Past Time. Cambridge: Cambridge University Press.

—. 1977. Household and Family in Past Time. Cambridge: Cambridge University Press.

Le Play, F. 1982. On Family, Work, and Social Exchange. Chicago: University of Chicago Press.

Lee, R.D. 1993. "Inverse projection and demographic fluctuations: A critical assessment of new methods." Pp. 7-28 in Old and New Methods in Historical Demography, edited by D.S. Reher and R. Schofield. Oxford: Clarendon Press.

Lee, R.D. and K.L. Kramer. 2002. "Children's economic roles in the Maya family life cycle: Cain, Caldwell, and Chayanov revisited." Population and Development Review 28(3):475-499.

Levin, I. 2004. ―Living apart together: A new family form.‖ Current Sociology 52(2):223-240.

Levitt, I. and C. Smout. 1984. "Farm worker's incomes in 1843." Pp. 156-187 in Farm Servants and Labour in Lowland Scotland, 1770-1914, edited by T.M. Devine. Edinburgh: John Donald.

Lewis, J.V.D. 1981. "Domestic labor intensity and the incorporation of Malian peasant farmers into localized descent groups." American Ethnologist 8(1):53-73.

248 Mitterauer, M. 1992. "Peasant and non-peasant forms in relation to the physical environment and the local economy." Journal of Family History 17(2):139-159.

Mitterauer, M. and R. Sieder. 1979. "The developmental process of domestic groups: Problems of reconstruction and possibilities of interpretation." Journal of Family History 4(3):257- 284.

Molin, G.D. 1990. "Family forms and domestic service in Southern Italy from the seventeenth to the nineteenth centuries." Journal of Family History 15(4):503-527.

Moring, B. 1999. "Land, labor, and love: Household arrangements in nineteenth century eastern Finland--cultural heritage or socio-economic structure?" The History of the Family 4(2):159-184.

Murtha, T.M., P.L. Johnson, J.W. Wood, S.A. Matthews, C.S. Sparks, and J.A. Jennings. 2008. "Historical demography, oral history, settlement archaeology, and landscape ecology: The North Orkney Population History Project." Presented at the annual meeting of the European Population Conference, Barcelona, Spain.

Nagata, M.L. 2005. "One of the family: Domestic service in early modern Japan." The History of the Family 10(4):355-365.

Naismith, R.J. 1985. Buildings of the Scottish Countryside. London: Victor Gollancz in association with Peter Crawley.

Netting, R.M. 1993. Smallholders, Householders: Farm Families and the Ecology of Intensive, Sustainable Agriculture. Stanford: Stanford University Press.

Netting, R.M.C., R.R. Wilk, and E.J. Arnould. 1984. "Introduction." Pp. xiii-xxxviii in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R.R. Wilk, and E.J. Arnould. Berkeley: University of California Press.

Neven, M. and C. Capron. 2000. Family Structures, Demography and Population: A Comparison of Societies in Asia and Europe. Liege: University of Liege.

Oeppen, J.E. 1993. "Generalized inverse projection." Pp. 29-39 in Old and new methods in historical demography, edited by D.S. Reher and R. Schofield. Oxford: Clarendon Press.

Orr, A. 1984. "Farm servants and farm labour in the Forth Valley and South-east Lowlands." Pp. 29-54 in Farm Servants and Labour in Lowland Scotland, 1770-1914, edited by T.M. Devine. Edinburgh: John Donald.

Otterbein, K.F. and C.S. Otterbein. 1977. "A stochastic process analysis of the developmental cycle of the Andros household." Ethnology 16(4):415-425.

Palsson, H. and P. Edwards. 1981. Orkneyinga Saga: The History of the Earls of Orkney. London: Penguin Books.

Perz, S.G. 2003. "Household demography and land use allocation among small farms in the Brazilian Amazon." Human Ecology Review 9(2):1-16.

249 Pinchbeck, I. 1969. Women Workers and the Industrial Revolution, 1750-1850. London: Virago.

Plakans, A. and C. Wetherell. 2005. "The Hajnal line and Eastern Europe." Pp. 105–128 in Marriage and the Family in Eurasia: Perspectives on the Hajnal Hypothesis, edited by T. Engelen and A.P. Wolf. Amsterdam: Aksant Academic.

Reay, B. 1996. Microhistories: Demography, Society and Culture in Rural England, 1800-1930. Cambridge: Cambridge University Press.

Redfield, R. 1989. The Little Community and Peasant Society and Culture. Chicago, IL: University of Chicago Press.

Reher, D.S. 1998. "Family ties in Western Europe: persistent contrasts." Population and Development Review 24(2):203-234.

Reyna, S.P. 1976. "The extending strategy: Regulation of the household dependency ratio." Journal of Anthropological Research 32(2):182-198.

Rodahl, K. 1989. The Physiology of Work. Philadelphia: Taylor and Francis.

Rudolph, R.L. 1992. "The European family and economy: central themes and issues." Journal of Family History 17(2):119-138.

Ruggles, S. 1987. Prolonged Connections: The Rise of the Extended Family in 19th Century England and America. Madison: University of Wisconsin Press.

—. 1992. "Migration, marriage, and mortality: Correcting sources of bias in English family reconstitutions." Population Studies 46:507-522.

—. 2009. "Reconsidering the Northwest European family system: Living arrangements of the aged in comparative historical perspective." Population and Development Review 35(2):249-273.

Sana, M. and D.S. Massey. 2005. ―Household composition, family migration, and community context: Migrant remittances in four countries.‖ Social Science Quarterly 86(2): 509-528.

Santley, R.S. and K.G. Hirth. 1993. ―Household studies in Western Mesoamerica.‖ Pp. 3-17 in Prehispanic Domestic Units in Western Mesoamerica: Studies of the Household, Compound, and Residence, edited by R.S. Santley and K.G. Hirth. Boca Raton: CRC Press.

Schrank, G. and D.G. Lockhart. 1995. An Orkney Estate: Improvements at Graemeshall, 1827- 1888. East Linton: Tuckwell Press.

Segalen, M. 1977. "The family cycle and household structure: five generations in a French village." Journal of Family History 2(3):223-236.

—. 1983. Love and Power in the Peasant Family. Rural France in the Nineteenth Century. Oxford: Blackwell.

250 —. 1984. "Nuclear is not independent: Organization of the household in the Pays Bigouden Sud in the nineteenth and twentieth centuries." Pp. 163-186 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

Shaw, F.J. 1980. The Northern and Western Islands of Scotland: Their economy and society in the seventeenth century. Edinburgh: John Donald.

Sieder, R. and M. Mitterauer. 1983. "The reconstruction of the family life course: theoretical problems and empirical results." Pp. 309-346 in Family Forms in Historic Europe, edited by R. Wall, J. Robin, and P. Laslett. Cambridge: Cambridge University Press.

Skinner, G.W. 1997. "Family systems and demographic processes." Pp. 53-95 in Anthropological Demography: Toward a New Synthesis, edited by D.I. Kertzer and T. Fricke. Chicago: University of Chicago.

Smith, R.M. 1984. "Families and their property in rural England, 1250-1800." Pp. 1-86 in Land, Kinship and Life-cycle, edited by R.M. Smith. Cambridge: Cambridge University Press.

Smock, P.J. 2000. ―Cohabitation in the United States: An appraisal of research themes, findings, and implications.‖ Annual Review of Sociology 26:1-20

Smout, T.C. 1986. A Century of the Scottish People, 1830-1950. New Haven: Yale University Press.

Snell, K.D.M. 1985. Annals of the Labouring Poor: Social Change and Agrarian England, 1660- 1900. Cambridge: Cambridge University Press.

Spagnoli, P.G. 1983. "Industrialization, proletarianization, and marriage: a reconsideration." Journal of Family History 8(3):230-247.

Sparks, C.S. 2007. Households, Land, and Labor: Population Dynamics in the Northern Orkney Islands, Scotland, 1851 to 2003. Ph.D. Dissertation, Anthropology and Population Research Institute, Pennsylvania State University.

—. 2009. "An application of the variable-r method to subpopulation growth in a 19th century agricultural population." Demographic Research 21:23-64.

StataCorp. 2009. Stata Statistical Software, Release 11. College Station, TX: StataCorp LP.

Stone, L. 1979. The Family, Sex and Marriage in England 1500-1800. New York: Penguin.

Szoltysek, M. 2007. "Central European household and family systems, and the ‗Hajnal– Mitterauer‘ line: The parish of Bujakow (18th–19th centuries)." The History of the Family 12(1):19-42.

—. 2008. "Three kinds of preindustrial household formation system in historical Eastern Europe: A challenge to spatial patterns of the European family." The History of the Family 13(3):223-257.

251 The National Archives of Scotland. 2003. Tracing Your Scottish Ancestors, The Official Guide: A Guide to Ancestry Research in the National Archives of Scotland. Edinburgh: Mercat Press.

Thomson, W.P.L. 1983. Kelp-making in Orkney. Stromness: The Orkney Press.

—. 2008a. The New History of Orkney. Edinburgh: Birlinn Ltd.

—. 2008b. Orkney Land and People. Kirkwall: The Orcadian Limited (Kirkwall Press).

Ulijaszek, S.J. 1995. Human Energetics in Biological Anthropology. Cambridge: Cambridge University Press.

Van de Walle, E. 1976. "Household dynamics in a Belgian village, 1847-1866." Journal of Family History 1(1):80-94.

Viazzo, P.P. 2005. "South of the Hajnal line: Italy and Southern Europe." Pp. 129-164 in Marriage and the Family in Eurasia: Perspectives on the Hajnal Hypothesis, edited by T. Engelen and A.P. Wolf. Amsterdam: Aksant.

Viazzo, P.P., M. Aime, and S. Allovio. 2005. "Crossing the boundary: Peasants, shepherds, and servants in a western Alpine community." The History of the Family 10(4):387-405.

Wakefield, D. 1998. Fenjia: Household Division and Inheritance in Qing and Republican China. Honolulu: University of Hawai'i Press.

Wall, R. 1983. "Introduction." Pp. 1-64 in Family Forms in Historic Europe, edited by R. Wall, J. Robin, and P. Laslett. Cambridge: Cambridge University Press.

—. 1986. "Work, welfare, and the family: An illustration of the adaptive family economy." Pp. 261-294 in The World We Have Gained: Histories of Population and Social Structure, edited by L. Bonfield, R.M. Smith, and K. Wrightson. New York: Basil Blackwell.

—. 2001. "The transformation of the European family across the centuries." Pp. 217-241 in Family History Revisited: Comparative Perspectives, edited by R. Wall, T.K. Hareven, and J. Ehmer. Newark: University of Delaware Press.

Watson, J. 1894. The Annals of a Quiet Valley, by a Country Parson. London: J. M. Dent.

Westray Buildings Preservation Trust. 2002. Fae Quoy tae Castle: An Orkney Island's Snapshot in Time. Kirkwall: Orcadian Limited.

Whittle, J. 2000. The Development of Agrarian Capitalism: Land and labour in Norfolk, 1440- 1580. New York: Oxford University Press, USA.

—. 2005. "Servants in rural England c. 1450-1650: hired work as a means of accumulating wealth and skills before marriage." Pp. 89-110 in The Marital Economy in Scandinavia and Britain, 1400-1900, edited by M. Agren and A.L. Erickson. Burlington, VT: Ashgate.

252 Whyte, I.D. 1989. "Population mobility in early modern Scotland." Pp. 37-58 in Scottish Society, 1500-1800, edited by R.A. Houston and I.D. Whyte. Cambridge: Cambridge University Press.

—. 2007. "Cumbrian village communities: Continuity and change, c. 1750-1850." Pp. 96-113 in The Self-Contained Village? The Social History of Rural Communities, 1250-1900, edited by C. Dyer. Hertfordshire: University of Hertfordshire Press.

Whyte, I.D. and K.A. Whyte. 1988. "The geographical mobility of women in early modern Scotland." Pp. 83-106 in Perspectives in Scottish Social History: Essays in Honour of Rosalind Mitchison, edited by L. Leneman. Aberdeen: Aberdeen University Press.

Wilk, R.R. 1984. "Households in process: Agricultural change and domestic transformation among the Kekchi Maya of Belize." Pp. 217-244 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

—. 1991. Household Ecology: Economic Change and Domestic Life Among the Kekchi Maya in Belize. Tucson: University of Arizona Press

Wilk, R.R. and R.M.C. Netting. 1984. "Households: Changing Forms and Functions." Pp. 1-28 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R.R. Wilk, and E.J. Arnould. Berkeley: University of California Press.

Winfield, F.E. 1985. Commuter Marriage: Living Together, Apart. New York: Basic Books.

Wolf, A.P. 1984. "Family life and the life cycle in rural China." Pp. 279-298 in Households: Comparative and Historical Studies of the Domestic Group, edited by R.M.C. Netting, R. Wilk, and E. Arnould. Berkeley: University of California Press.

Wolf, E.R. 1966. Peasants. Englewood Cliffs: Prentice-Hall.

Wood, J.W. Forthcoming. Population, Food, and Traditional Farming: A New Synthesis.

Wrigley, E.A. 1994. "The effect of migration on the estimation of marriage age in family reconstitution studies." Population Studies 48(1):81-97.

Wrigley, E.A., R.S. Schofield, R.S. Davies, J.E. Oeppen, and C. Wilson. 1997. English Population History from Family Reconstitution. Cambridge: Cambridge University Press.

Yanagisako, S.J. 1979. "Family and household: The analysis of domestic groups." Annual Review of Anthropology 8(1):161-205.

VITA

Julia Anne Jennings

Education Ph.D. Anthropology and Demography, minor Statistics The Pennsylvania State University, 2010 M.A. Anthropology and Demography, The Pennsylvania State University, 2009 B.S. Anthropology, summa cum laude, Tulane University, 2005 B.A. Medieval Studies, minor Philosophy, cum laude, Tulane University, 2005

Experience 2010 NICHD-NRSA Postdoctoral Fellowship, Carolina Population Center, University of North Carolina at Chapel Hill 2005-10 Predoctoral Trainee, Teaching/Research Assistant, Graduate Fellow Department of Anthropology, Population Research Institute, The Pennsylvania State University

Selected Research & Presentations 2010 Jennings, JA, Hacker, JD, and Sullivan, AR. Intergenerational Transmission of Reproductive Behavior during the Onset of the Fertility Transition in the United States: New Evidence from the Utah Population Database. Journal of Interdisciplinary History. [In Review] 2010 Jennings, JA. Household Predictors of the Presence of Life Cycle Servants in Orkney, Scotland, 1851-1901. Paper presented at the Meetings of the Population Association of America, Dallas TX. 2009 Jennings JA, Sparks CS, Wood JW, Johnson PL, Murtha TM, and Matthews S. Formation and Dissolution of Extended-family Households in Northern Orkney, Scotland, 1851-1901. Paper presented at the IUSSP conference in Marrakech, Morocco. 2008 Murtha TM, Johnson PL, Wood JW, Matthews SA, and Jennings JA. Historical Demography, Oral History, Settlement Archaeology, and Landscape Ecology: The North Orkney Population History Project. Paper presented at the European Population Conference, Barcelona, Spain. 2008 Jennings JA, and Wood JW. Multi-Census Methods for Anthropological Demography: Applications to the Northern Islands of Orkney, Scotland. Poster presented at the Meetings of the Human Biology Association, Columbus, OH.

Fellowships & Grants 2009 Penn State RGSO Travel Grant 2007-9 NICHD Pre-doctoral Traineeship in Demography, Population Research Institute 2008 Sanders Graduate Research Award, Department of Anthropology, Penn State 2008 German Academic Exchange Service (DAAD) Travel Grant 2007 ICPSR, University of Michigan, summer stipend and tuition 2007 Baker and Hill Fellowship, Department of Anthropology, Pennsylvania State University 2005 University Graduate Fellowship, College of Liberal Arts, Pennsylvania State University