Time and space in historical demography: some case studies using Dutch micro-data

(1)

University of Groningen

Time and space in historical demography

Ekamper, Peter

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.

Document Version

Publisher's PDF, also known as Version of record

Publication date: 2019

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

Ekamper, P. (2019). Time and space in historical demography: some case studies using Dutch micro-data. University of Groningen.

Copyright

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

(2)

Time and space in historical

demography

(3)

Cover illustration: Map of Groningæ et Omlandiæ – De Provincie van Stadt en

Lande by Ludolpho Tjardæ â Starckenburg and Nicolaus Viſscher (around 1681)

No part of this thesis may be reproduced, stored in a retrieval system, or transmitted in any form by any means without the prior written permission of the author.

(4)

Time and space in historical

demography

Some case studies using Dutch micro-data

PhD thesis

to obtain the degree of PhD at the University of Groningen

on the authority of the Rector Magnificus prof. E. Sterken

and in accordance with the decision by the College of Deans. This thesis will be defended in public on

Monday 1 July 2019 at 15:45 hours

by

Peter Ekamper

born on 15 March 1964

(5)

Supervisors

Prof. L.J.G. van Wissen Prof. C.J.I.M. Henkens

Assessment committee

Prof. H.A.J. Bras Prof. M.G.J. Duijvendak Prof. F. Janssen

(6)

1. Introduction

Novel approaches to time and space that are gaining ground in the field of demography have been coming up at a time in which historical demographers are getting access to new, much larger and more fine-grained data that stretch over long periods of time and cover vast areas. The methodological advances have given historical demographers new sets of tools to examine received ideas and pose new questions. New methods and theoretical approaches that have been developed and new and better databases enable historical demographers to take up new and old research questions from within and outside the traditional field, offering excellent opportunities for the further development of historical demography (Southall, 2014). This thesis presents four case studies of historical demographic topics on the Netherlands using these newly available sources of historical micro-level demo-graphic data. The case studies aim to provide a varied picture of the possibilities of the use of these historical Dutch micro-level demographic data in studying demo-graphic events in a spatial and temporal context. The case studies cover a mix of regions and periods and use different microdata sources. The first case study deals with the impact of heat waves and cold spells on mortality in the nineteenth and the first half of the twentieth century in the Netherlands using individual death records. The second case study focuses on the changing geography of the Dutch marriage market in the nineteenth and early twentieth century using individual marriage records. The third case study is about the effects of religion, social class, and space on infant mortality in mid-nineteenth century Amsterdam using population register records. The last case study is on war-related mortality among the civilian population in the Netherlands during the last stage of World War II using individual causes of death registry data.

Before presenting the results of the case studies in the next chapters, the remainder of this chapter will first address the reawakened interest and importance of space and place in historical demography. Next, some developments and examples

(11)

will be presented of research in areas of historical demographic topics particularly suited for application of spatio-temporal analyses; especially the study of mortality (crises) in time and space, like the effects of epidemic diseases (and their diffusion) and climatic changes on mortality. The subsequent section will discuss the additional role and importance in historical demographic research of spatial differences in the socio-historical context, both at the individual and aggregate level. This chapter concludes with an overview of some of the most important historical micro-level demographic datasets and historical spatial context data, followed by an outline of the next chapters on the case studies.

Space and place in historical demography

For a long time demographic research mostly consisted of analyses of differences in population trends and demographic change between geographic areas. However, from the 1960s on demography focused more and more on the individual, the family and the household and mostly ignored the geographic dimension (Voss, 2007). In the late 1980s and early 1990s space and place again became key concepts in demography. Voss (2007) called this growing interest in the spatial context of social processes a “reawakening to matters of space and place”, the growing realization that the spatial context in which individuals are embedded has an impact on their demographic behaviour. At the same time, it became much easier to bridge the gap between micro-demographic data and macro-demographic geographical aggregates. First, there is a growing availability of fine-grained spatial data, linked to geographic coordinates and their demographic, social and economic characteristics. Moreover, the availability of advanced computing facilities increased tremendously. Beyond that there is the ongoing development of geographical information systems (GIS), other specialized software for mapping and integrating spatial data, and the development of tools and techniques for the specification and estimation of statistical models based on spatial data (Anselin, 1988; De Smith et al., 2007; Scholten, Van de Velde and Van Manen, 2009; Stillwell and Clarke; 2003). Additionally, these tools and techniques are increasingly becoming available in general-purpose statistical software packages (Bivand, Pebesma, and Gómez-Rubio, 2013; Bivand and Piras, 2015; Pisati; 2012). All this provided the opportunity to study demographic processes at the level of the individual person embedded in a spatial context, leading to novel approaches to time and space in demographic research (Entwisle, 2007; Howell and Porter, 2013; Howell, Porter, and Matthews, 2016; Matthews and Parker, 2013; Voss, 2007; Wachter, 2005).

(12)

3

The effects of this new approach are visible in historical demography as well. Space and time were central concepts in historical demography from the beginning and the combination of space and time is a key element in historical studies of populations (Gutmann et al., 2011). However, despite decades of debate about for instance the geography of historical family and household composition, there has been little spatial analysis; time has likewise often been absent from historical (family) demography (Ruggles, 2012). But the study of population history now has shifted from studying demographic regimes and large-scale processes to analysing longitudinal micro-data (Alter, Mandemakers, and Gutmann, 2009). Longitudinal demographic micro-data enables historical demographers to analyse processes of demographic change at the individual and family level; processes that span life times and can even be followed across generations. Besides the new theoretical approach, the innovative methods of spatial-temporal analysis have made their way into historical-demographic research as well (Gregory, DeBats, and Lafreniere, 2018; Gregory and Geddes, 2014; Gutmann et al., 2011). The prospects for historical demographic studies in which individual level data are integrated with combined space and time information have been affected by the same factors that have played a role in demographic studies in general. New and improved historical datasets have been constructed. In the last decade there has been enormous growth in individual-level historical population datasets (Ruggles, 2014), primarily consisting of digitally transcribed nineteenth and early twentieth century census data and vital population register data. See, for instance, Song and Campbell (2017) and Ruggles (2014) for overviews of major historical demographic micro-level data infrastructure projects worldwide. These datasets are increasingly matched with information on the geographical location of the spatial units in which individuals spent their lives as well as with information on the characteristics of these spatial units. However, there are but few studies that have constructed or used geocoded longitudinal demographic data (Hedefalk, Harrie, and Svensson, 2015).

Mortality (crises) in time and space

The study of mortality – and especially of mortality crises – is particularly suited for the application of spatio-temporal analysis. Prior to the industrial revolution, the mortality of human populations was often characterized by dramatic, unpredictable increases due to demographic shock events or crises like epidemics, famines, wars and displacement of people, which could wipe out a large part of the population of a community within a very brief period (Livi Bacci, 2001). One of the earliest and most

(13)

famous examples of the value of spatial and temporal analysis in studies in the field of demography was provided by the English physician and medical doctor John Snow during the 1854 London cholera epidemic, revealing a clustering of deaths around the Broad Street water pump (Snow, 1855). And still today the diffusion of diseases and deaths, at the level of individuals, communities or other geographic units is inherently suited for a spatial and temporal analysis. Historical demographers have always shown a keen interest in the study of epidemic disease mortality, such as cholera, smallpox, typhus and influenza, the identification of the most vulnerable population groups, and the seasonal distribution of epidemic deaths. The geographical context is fundamental to spatial disease spread. Important factors in this context are physical proximity, heterogeneity of environmental and infrastructural characteristics, spatial differences in behaviour and characteristics of individuals and groups to which they belong, as well as spatio-political decision making (Kennedy, Curtis, and Curtis, 2015). A number of studies on the spatial spread of epidemics particularly in urban contexts used individual-level data linked to Geographical Information Systems. These studies analyse and model the spatial contamination and diffusion pathways of epidemics between individuals, households and families in relation to the local spatial environment. See for instance Fornasin, Breschi, and Manfredini (2011) and Skillnäs (1999) on nineteenth century cholera outbreaks in respectively Udine, Italy and Linkoping, Sweden; Séguy, et al.(2012) on

an 18th_{century probable whooping cough epidemic in Martigues, France; and Cain}

(2004) on a nineteenth century smallpox epidemic in Sheffield, England. The key questions central to these studies are how diseases and epidemics spread geographically over time within local urban environments and to what extent these patterns can be explained by locally varying conditions, like sanitation and infrastructure. Some of these studies have applied sophisticated diffusion models and GIS-based research linking epidemics for example to water supply and sewage systems. Studies of diffusion patterns of epidemic diseases not within but between spatial units have become very popular as well. For instance Smallman-Raynor, Johnson and Cliff (2002) and Valleron et al. (2010) have explored the origin, spread and impact of the 1889 and 1918-19 influenza epidemics. These studies, making use of advanced spatial statistical analysis, established clearly defined processes of spatial contagion, contrary to the contemporary observations. However, problems inherent to this kind of studies are numerous: the digitalization, vectorization and georeferencing of information, the identification of individuals and their location in time and space, the cross-linking with spatial information on their houses and neighbourhoods are all very complex processes (Cliff and Haggett, 1988). These types of problems become even more prominent when studying circumstances where

(14)

5

data is often unreliable or missing such as demographic shock events like famine, climatic disasters, wars and displacement of people. Although individual-level data are also becoming more and more available on such disastrous events, such data is in general still too scarce, incomplete or insufficient to reconstruct mortality patterns at the individual level. Studies in these topics therefore usually focus on data at aggregated spatial levels. With respect to famines, for instance, Fotheringham, Kelly and Charlton (2013) studied the Irish famine of the 1840s, making use of advanced spatial statistical analysis using aggregate data at the spatial level of census enumeration districts. Studying the Ukrainian famine of 1932-1934, Wolowyna et al. (2016) used reconstructed data at the county level.

With respect to the impact of climate on mortality there has been an increased recognition of the potential effects of long-term climate change (Haines et al., 2006; Keatinge and Donaldson, 2004). To determine whether changes in mortality might be expected to correspond with future changes in climate, empirical observations of the relation between short-term variations in weather and mortality over a long time period are indispensable (Haines et al., 2006). From a historical perspective it is interesting to study how societies in the past coped with environmental shocks such as extremely high temperatures and whether or not they were able to restrict the ensuing effects on mortality (Galloway, 1994). Studying the differences in vulnerability to environmental stress by social class, age and sex, gives us information on the conditions under which these groups lived and on the way their conditions changed over time (Bengtsson, Campbell, and Lee, 2004). However, the number of studies in which the relation between weather variations and mortality over a long period of time is studied is rather limited. They rarely cover periods long enough to encompass major economic, demographic or epidemiological transitions (Carson et al., 2006). Historical studies of the relationship between weather and mortality are usually based on rather crude weather and mortality indicators. Mortality is often available only for the population as a whole, without de-composition by sex or age. Data on the numbers of deaths are often given per month only; see for example Galloway (1988; 1994). With more refined data, for example daily death counts, by age, sex and social class, it is possible to observe patterns of heat or cold-related mortality. This kind of data has become increasingly available in many European countries over long periods of time and for larger regions. Databases with historical series of homogenized temperature measures are developing as well. New methods have been developed to study this kind of data (Eilers et al., 2008; Rau, 2007), which enables us to study in a comparative way the long-term and area-specific relationship between weather conditions and demographic changes. In historical populations where infectious diseases were the dominant cause of death,

(15)

the season of birth, the age at which weaning started, seasonal temperature fluctuations and the quality of drinking water together determined to a large extent the within-yearly fluctuations in the numbers of deaths. A study by Carson et al. (2006) used twentieth century micro- and macro-data to analyse temperature-related weekly mortality in London. Dalla-Zuanna and Rosina (2011) and Derosas (2009) combined nineteenth century micro-level data on neonatal mortality with daily data on temperatures for northeastern Italy. Åström et al. (2013) studied the short-term effects of extreme hot and cold weather on mortality in Stockholm using twentieth century daily mortality data. Petkova, Gasparrini and Kinney (2014) analysed heat and mortality in New York City in the twentieth century using daily temperature and mortality data. Ekamper et al. (2009) studied the changing impact of cold spells and heat waves on mortality using Dutch individual death records with daily temperature in the southwest of the Netherlands for the twentieth century and second half of the nineteenth century. Jennings and Clark extended the analysis of the effects of climate to other demographic events. They analysed the effects of climate variability on migration (Jennings and Clark, 2015) and on marriage (Jennings and Clark, 2017) for the second half of the nineteenth century and the first half of the twentieth century using individual-level demographic sample data for the Netherlands.

Spatial socio-historical context

An important issue in the study of mortality, or any other demographic event, is the role of spatial variety in the socio-historical context, both at the individual and aggregate levels. Time and space have a central role in the increasing number of mortality studies that take into account not only the individual characteristics of persons, but also the historical spatial context in which socio-demographic groups of individuals are embedded (Hummer, Rogers, and Eberstein, 1998). Examples are studies showing that living in an area characterized by poor socio-economic conditions had negative effects on health for people with both a high and a low socio-economic status. In historical mortality studies, this has frequently been suggested as an explanation for observed trends in mortality differences between individuals. Smith (1991) for example has argued that in a mortality regime dominated by variation in the incidence of infectious diseases the location in a spatially-structured disease environment mattered more for mortality variation between individuals than the characteristics of individuals, due to the lack of residential segregation and the frequent use of servants. Garrett et al. (2001) observed that in early twentieth

(16)

7

century England and Wales, infant mortality in certain occupational groups differed according to the socio-economic structure of the area in which they lived. A growing number of studies now indeed include indicators of socio-economic position both at the individual and the community level, making it possible to study demographic outcomes as an effect of the characteristics of the individuals living there and of those of the location. This requires detailed data at the individual and the aggregate level and at the same time the use of models that allow us to take into account the spatial dependency of the subjects and areas to separately estimate the effects of the spatial context from that of the characteristics of the people living there. Fortunately, for a growing number of places, micro-level spatial data have become available. Recent examples in historical demography of such studies include Thornton and Olson (2011) on mortality in late nineteenth-century Montreal; Sawchuck, Tripp and Melnychenko (2013) on mortality in late nineteenth-century Gibraltar; Xu, Logan and Short (2014) on child mortality in 1880 Newark, New Jersey; Jaadla and Puur (2016) on infant mortality in 1897-1900 in Tartu, Estonia; and Connor (2017) on child mortality in the early twentieth century in Dublin, Ireland. Place is conceptualized as a source of health access or exposure and a variety of indicators is used to determine its effects: like socio-economic resources, altitude, population density, ethnic diversity, and environmental hazards such as the presence of tap water and sewing system (see for instance Beemer, Anderton and Hautaniemi Leonard, 2005; Jaadla and Puur, 2016; Thornton and Olsen, 2011). Numerous problems arise, however, when studying in an integrated way the roles of neighbourhood and spatial variations in the health of individuals: the choice of the data aggregation level, the identification of real-life neighbourhood boundaries that determine the everyday social and physical interaction among residents, determining spatial proximity, etc.

Historical micro-level demographic data

The case studies presented in this thesis use historical micro-level demographic data extensively. In the last decade there has been enormous growth in historical micro-level demographic datasets, and the quantity of accessible historical microdata is expected to keep growing (Ruggles, 2014). Historical sources of demographic microdata include censuses, population registers, household registers, family genealogies, and church and vital records. Historical census microdata for several European countries and regions are collected and made available through the Mosaic project (Szołtysek and Gruber, 2016). Ruggles (2012) and, more extensively, Song

(17)

and Campbell (2017) provide worldwide overviews of historical (and contemporary) data sources, especially focused on linked datasets, including census records, population registers, parish registers, vital records, and family genealogies – covering periods from the early seventeenth to the twentieth century. Typically, these datasets include data on demographic events such as fertility, mortality, marriage, and migration, as well as indicators of socioeconomic status like occupation (Song and Campbell, 2017). Some examples of these datasets are the Integrated Public Use Microdata Series (IPUMS) Linked Representative Samples based on the United States census records, covering 500,000 individuals over the period 1850-1930; the Swedish POPLINK Demographic Database, covering 350,000 individuals over the period 1680-1950; the Norwegian Historical Population Registers (HPR) , covering 9.7 million individuals over the period 1801-1964; the Canadian BALSAC population database, covering 5 million individuals over the period 1621-1965; and the Dutch LINKing System for historical family reconstruction (LINKS/GENLIAS), covering 11 million individuals who died in the period 1811-1965.

The Dutch LINKS/GENLIAS database project aims at reconstructing all nineteenth and early twentieth century families in the Netherlands. The database includes and links micro-level data from birth, death, and marriage civil certificates for seven of the Dutch provinces: Groningen, Drenthe, Overijssel, Gelderland, North-Holland, Zeeland and Limburg, which makes the dataset a unique source for historical demographic research on the Netherlands in the nineteenth and early twentieth century. However, the dataset is still in development and the data do not (yet) cover the whole of the country. Another important Dutch historical micro-level demographic dataset is the Historical Sample of the Netherlands (HSN); a representative sample of about 78,000 people born in the Netherlands during the period 1812-1922 (Mandemakers, 2002). The HSN contains data on individual life-courses including data on demographic events (like age at marriage, number of children born), religious affiliation, occupation, birth place, literacy, social network and migration history. Besides these national initiatives, there is an increase of historical micro-level local and regional demographic datasets provided as open data by Dutch local and regional archive institutions, like the city archives of Delft, Leiden, and Rotterdam, regional archives of Alkmaar and North Brabant, and provincial archives of Drenthe, Friesland, Groningen, Limburg, Overijssel, and Zeeland. These datasets often not only include digitized and transcribed nineteenth and early twentieth century birth, death, and marriage certificates from the civil registration, but also pre-nineteenth century church and parish registers of baptisms, marriages, and burials. Some of the latter registers even date back to the sixteenth century.

(18)

9 Historical spatial context data

To analyse demographic data in a spatial context, these data have to be linked to geographical locations. In the case studies presented in this thesis, micro-level demographic data have been linked to geographical locations at different spatial levels depending on the most detailed spatial level available for the study topic, varying from municipalities and weather stations at a more macro-level to residential addresses at the micro-level. Marriage data and war-related mortality data were available at the spatial level of municipalities. Climatic data were restricted to the level of weather stations. Infant mortality data were available at the level of residential addresses. Within the field of historical demography, many studies have linked longitudinal micro-level (individual) or macro-level (aggregated) demographic data to the spatial context on a macro-level, for instance at the level of counties, municipalities, neighbourhoods, parishes, or wards. Studies like Atkinson et al. (2017), Fotheringham, Kelly and Charlton (2013), and Haines and Hacker (2011) linked aggregated demographic data in this way. Jaadla and Reid (2017) and Tinghög et al. (2011), for instance, similarly linked both aggregate and individual demo-graphic data. Studies linking micro-level demodemo-graphic data to the micro-level spatial context at the level of for instance residential addresses, dwellings, or properties are also becoming more common (see for example Cain, 2004; Ekamper, 2010; Fornasin, Breschi, and Manfredini, 2011; Séguy, et al., 2012; Skillnäs, 1999). However, these studies are seldom longitudinal because there are few historical longitudinal demographic datasets that geocode individuals to a precise spatial location, such as property units, parcel boundaries, buildings, or residential addresses over long time periods (Hedefalk et al., 2017). Some recent studies that do link longitudinal micro-level demographic datasets to the micro-level spatial context are Hedefalk, Harrie and Svensson (2015) and Villareal et al. (2014). Hedefalk, Harrie and Svensson (2015) are using historical data from the Scanian Economic Demographic Database (SEDD) to link approximately 53,000 individuals in five Swedish rural parishes for the period 1813-1914 to the property units in which they had lived. Villareal et al. (2014) created a historical urban ecological dataset covering seven U.S. cities during the period 1830-1930 combing ward level, street level, and residential address level data.

Linking micro- and macro-level geographical units is not without problems. Particularly when exploring long-term spatio-temporal change using data published for administrative geographical units, analyses often need to be adjusted for boundary changes over time (Gregory, 2008 and 2009). For instance in the Netherlands the number and territories of the administrative units of municipalities

(19)

have changed drastically over the years (Van der Meer and Boonstra, 2006), from more than 1,200 in the beginning of the nineteenth century to less than 400 since 2015. Another complicated issue is the transcription, geocoding, geolocating and georeferencing of geographical names, and even more for historical geographical names (Reba, Reitsma, and Seto, 2016). Helpful sources for linking Dutch nineteenth and twentieth century macro-level spatial data are Van der Meer and Boonstra (2006) for municipal boundary changes and Huijsmans (2013) for geocoding and georeferencing Dutch hamlet, village, town, city, and municipality place names.

The corresponding spatial context to micro-level demographic data is that of the place of living or other places of life course events of the individual or household, viz. dwellings and other residential buildings. These can be identified by residential street addresses and house numbering, or the geographical latitude and longitude coordinates of the residential location. However, particularly in a longitudinal context geocoding, geolocating and georeferencing of such data can be very complex due to the renumbering of historical addresses over time, inaccurate historical mapping, incompleteness of sources, geographic name changes, multiple geographic name spellings etc. (Ekamper, 2010; Hedefalk, Harrie, and Svensson, 2015; Villareal et al., 2014). Usually the most accurate and most detailed historical maps available are cartographic records of property ownership, viz. cadastral maps. Cadastral surveying, registration and mapping has a long tradition in countries like the Netherlands and Sweden, with very early cadastral maps dating back from the beginning of the seventeenth century (Kain and Baigent, 1992). Only during the nineteenth century did most countries start to produce really accurate large-scale cadastral maps (Kain and Baigent, 1992). The modern Dutch cadastre originates from the period of French annexation by Napoleon from 1810 to 1813. The cadastral surveying resulted in a production of almost eighteen thousand large-scale maps in a period of almost twenty years. Parcels were numbered on the map corresponding with the number scheme of the cadastral ledgers which contain name, place of residence, and occupation of the land owners, as well as type of land use, area, and taxable value classification. The surveying was completed in 1831 and the Dutch Cadastre became formally operative in 1832. Through a joint effort of the Dutch national and regional archives the almost eighteen thousand maps and the accompanying 150,000 pages of the cadastral ledgers of the Dutch Cadastre of 1832 were digitised in the form of digital high-resolution raster images in the period 2001-2003 (Ekamper, 2010). The Fryske Akademy (2018) in the Netherlands, started in 1997 vectorising and georeferencing the cadastral maps of the province of Friesland, gradually expanding this provincial historical GIS released in 2005 to a national Dutch Historical GIS (HISGIS) in development both in space and time; in 2018

(20)

11

covering the 1832 situation in the provinces of Drenthe, Friesland, Gelderland, Groningen, Overijssel, Utrecht, and Zeeland fully, the province of North Brabant partly, and a few cities in the rest of the country, among which Amsterdam and Rotterdam.

Linking Dutch micro-level demographic data to these largescale Dutch cadastral maps is a complicated and time-intensive matter (Ekamper, 2010). Street addresses and house numbering used in micro-level demographic sources like population censuses and population registers are different from the cadastral numbering. Census and population register house numbering as well as cadastral numbering also changed over time. The city of Amsterdam for instance, introduced a house numbering system in 1796. However, at the same time the previous used tax numbering system of the houses (verpondingsnummers) was still kept in use. In 1853 the city decided to change to a new house numbering system and in 1875 again to the numbering system still in use. Mid-nineteenth century Amsterdam population register records in particular show inconsistent use of the old and new numbering systems.

Historical demographic data and spatial contexts used in this study In the next chapters several Dutch sources of historical demographic micro-level data will be used. The first micro-level data source used is the Dutch LINKS/GENLIAS database roughly covering the period 1811-1920s. The data used from this dataset are the micro-level data from both the death records and the marriage records. The historical death records dataset provides date of death, age, gender, marital status, place of birth, place of living, place of death, and occupation (or, in case of children, occupation of the parents) of the deceased. The historical marriage records dataset provides for each marriage the date and place of marriage, and for the bride and the groom the age, place of birth, place of living, and occupation (and occupation of the parents). The most detailed spatial context level available in this dataset is the place, viz. municipality, of the demographic events, like place of marriage, place of death, and place of birth. Although some municipalities, particularly larger cities and towns, recorded residential addresses in the original certificates as well, that information has not been digitized and would not have been available for all other municipalities anyhow.

The second micro-level data source used is the Amsterdam population register 1851-1853. The Amsterdam population register provides for each registered house-hold member the residential address, date and place of birth, marital state,

(21)

occupation, religion, and if applicable, date of death, date of moving in and date of moving out. The Amsterdam population register data have been supplemented and updated with missing birth data from the birth certificates and stillbirths from the death certificates of the civil registration. Since the municipality of Amsterdam registered residential addresses both in the population register and the birth, death and marriage certificates from the civil registry, the most detailed spatial context level available in this dataset is the spatial micro-level of the residential address. This allows combining the micro-level demographic data with the micro-level spatial data from the digitized version of the nineteenth century Dutch Cadastre as implemented in the Historical GIS Amsterdam.

The third micro-level data source used is the cause of death registry

(Doods-oorzakenstatistiek) from Statistics Netherlands, an electronic database that includes

registered deaths in the Netherlands from January 1936 to the current date. The dataset contains individual death records with data on date of death, age at death, gender, cause of death, place of death, and place of residence. Due to privacy requirements of Statistics Netherlands, the most detailed spatial context level available in this dataset is the place (municipality) of death.

Outline of the book

The next chapters of the book, chapters 2 to 5, present the four case studies of historical demographic topics on the Netherlands. These chapters are written as separate articles, all of which have already been published.

Chapter 2 deals with the impact of heat waves and cold spells on mortality in the nineteenth and the first half of the twentieth century in the Netherlands. The study uses Dutch micro-level death records data from the LINKS/GENLIAS database in relation to regional daily temperature data for the period 1855-1950 for four of the then eleven Dutch provinces (the provinces of Drenthe, Gelderland, Groningen, and Utrecht). Negative binomial regression models were used to analyse whether the effect of extreme heat and cold varied by province, age, sex and social class, and to analyse the changes in vulnerability to temperature fluctuations.

Chapter 3 deals with the changing geography of the marriage market in the nineteenth and early twentieth century in the Netherlands. The study uses the Dutch micro-level marriage records data of the LINKS/GENLIAS database for the period 1812-1922 for five Dutch provinces (the provinces of Drenthe, Gelderland, Groningen, Overijssel, and Utrecht). The geographical distances between the birth-places of spouses, as indicated in marriage certificates, were used to measure

(22)

13

increasing spatial interaction and widening geographic horizons over time. Various descriptive measures of the change in distances between spouses and in directional preferences are presented, as well as an analysis of the changing relationship between social position and geographic horizon.

Chapter 4 focuses on the effects of religion, social class, and space on infant mortality in mid-nineteenth century Amsterdam. This study uses Dutch micro-level demographic data from birth, death, and population records for infants born in the city of Amsterdam in 1851 linked to micro-level spatial locations from historical cadastral maps and other spatial data. Cox's proportional hazards models were used to analyse the effect of various sociodemographic characteristics, residential environment, water supply, and health care variables on infant mortality and stillbirth.

Chapter 5 deals with war-related mortality among civilians during the last stage of World War II in the Netherlands. The study uses Dutch micro-level data on monthly mortality and causes of death from Statistics Netherlands covering the period 1944-1947 for the entire country. Despite there being several estimates particularly for famine-related deaths in the west of the Netherlands during the last stage of World War II, no such information existed for war-related excess mortality among the civilian population from other areas of the country. The previously unavailable mortality data allow estimating the number of war-related excess deaths during the last stage of the war in the whole country by age group, sex, cause of death, and region. The study uses a seasonal-adjusted mortality model combined with a difference-in-difference approach to estimate the number of excess deaths in the period between January 1944 and July 1945.

(23)

(24)

15

2. Heatwaves and cold spells and their

effect on mortality

An analysis of micro-data for the Netherlands in

the nineteenth and twentieth centuries

1

For many centuries, Hippocrates’ thoughts on the role that weather conditions play on human health were a source of inspiration for physicians. Following the ideas that Hippocrates had formulated in his On Airs, Waters, and Places, physicians studied the ways in which the natural conditions of a country affected the appearance and virulence of diseases. When vital registration systems were introduced in Europe in the early nineteenth century and large-scale mortality data became available, empirical studies on the effect of periods of extreme heat or cold became possible for the first time. Doctors were now able to identify the negative health effects of heat and cold. For the Netherlands we can document the interest in this topic not only with statistical studies and official reports but also with literary sources and autobiographies, personal documents, newspaper articles and iconographic sources.

In 1868 summer temperatures in the Netherlands reached extremely high levels. With the help of locally collected data doctors studied the unusually strong increase in infant mortality (Gedeputeerde Staten Zeeland, 1869; Godefroi, 1869). A leading Dutch hygienist, Casper Pieter Pous Koolhaas (1831-1893), stated that in many places, in 1868 “more often than in many other summers, in the summer of this year an infant’s body has been carried to the grave”. The main reason for this extreme summer mortality was the kind of food supplied to children. Artificial feeding of children became even more difficult than it already was under normal circumstances. During hot weather, foods such as milk and bread porridge under-went “a slight change, and in this process start to decay […]. To dilute the milk, or to prepare other foods or drinks, water is used, and if one considers how poor the water is in some places as a consequence of the heat and drought, then one has another

1_{This chapter is a slightly modified version of: Ekamper, P., Poppel, F. van, Duin, C. van, and} Mandemakers, K. (2011), Heat waves and cold spells and their effect on mortality: An analysis of micro-data for the Netherlands in the nineteenth and twentieth centuries. Annales de Démographie

(25)

reason for the harms that are caused in particular during hot summers by artificial feeding of infants”. “The poorer the water used, and the sooner and broader the decay of the food, the higher too is the risk of uncleanness of the teats of the bottles in which one hands the food to the children.” Pous Koolhaas argued that in hot summers the mortality of children of the poor increased more strongly than among children of the well-to-do. “It is among these families in whose dwellings, sooner than in the houses of the rich, the bad air manifests itself; in which, be it out of ignorance, negligence or frugality, more often bad or badly prepared food is given to the children, or what has been left from an earlier day [...]. Lack of discernment, unfamiliarity with the need for it also leads in these families to a less-than-required care for the complete purity of bottles and other utensils in which the child’s food is kept and which is so extremely important, particularly in hot weather. The milk that is bought by those people for whom spending a few more cents is a question of high importance will generally be poorer in quality than the one that can be supplied by those who wish to buy good stuff, even if that costs a bit more; and those who buy the bad and often-old milk, very frequently will have more trouble in trying to preserve it from decay.” (Pous Koolhaas, 1869).

The heat wave of 1911 drew even more attention than the one of 1868, as the latter’s death toll was indeed considerable. “Enormous loss of human lives”, in particular among infants, “a sad phenomenon, unparalleled in the statistics of recent years”, “a massacre” especially in the countryside and less so in the larger towns “with their good drinking water and controlled milk stations”– in these terms the Dutch press described in November 1911 the effect of the heat (De statistiek van den loop der bevolking, 1911). Medical doctors studied its consequences in detail, especially for infants. A survey had been set up in The Hague which followed for several years all children born in that city in 1908. The researchers had ample opportunity to pay attention to the effect of temperature on the sampled infants (Gezondheidscommissie 's-Gravenhage, 1913, 72-89). They stressed that the effect of heat periods was not only due to the rise in temperature as such but also depended on the duration of the heat period. The authors observed that mortality had increased not only for children aged one month or older but also in the first month of life. It appeared that after the extreme mortality in the summer months the death risks for infants in November and December were lower than in normal years. Detailed daily temperature and morbidity and mortality data were analyzed to examine the short- and long-term effects of temperature on illness and death. The authors were able to show that the number of days elapsing between the heat peak and the mortality peak shortened when the heat period kept up and temperatures rose further. Heynsius van den Berg (1912) found out through weekly data on

(26)

17

numbers of infant deaths that the larger cities had withstood the summer of 1911 relatively well, a finding that he explained by the hygienic measures that had been taken there in the recent past. He observed no direct adverse effect of heat on infants; it was only after a series of days with extremely high temperatures that the situation for infants became unbearable. This was a consequence of the fact that outdoor as well as indoor temperatures reached very high levels. Data for Amsterdam also shows the effect of increases in indoor temperature (De Lange, 1913). In September 1911, when the heat had already disappeared but houses were still overheated, high mortality peaks could still be found. De Lange observed that the heat had less of an effect on children younger than one month. She credits this to very young children having relatively high rates of heat loss and the fact that many of these children were still being breastfed. Many deceased children had lived in single-room dwellings, and quite often the cooking and laundry was done in the single-room where the infant stayed during the daytime. Another problem was that in periods of extreme heat thirsty children received solid food or sour milk, and many infants were dressed with an excess of clothes (De Lange, 1913).

Contemporary observers also noticed the mortality-increasing effect of extreme winter temperatures. A case in point was the winter of 1890-1891, during which the agrarian population in particular, and among them the poor in the first place, suffered. An anonymous letter to the editor of a local newspaper (Het

Nieuwsblad) on January 10th_{, 1891 described the situation in a region called the}

Hoeksche Waard, an island slightly north of the province of Zeeland. “The barren winter which quite unexpectedly holds sway with implacably harshness takes a heavy toll among the well-to-do, but how much more does it take out of our destitute human beings. […] Go and visit our poor and abide a few moments at the bedside of numbed old people; look at the chilly, shivering children, covered with rags, yearning for a nourishing meal or a warming fire. Look at those toddlers, sleeping under pieces of rugs, floor mats, shredded clothes, or other things that a caring parental hand has been able to find as cover; see with your own eyes that in our midst people can be found who are less well-off than animals in the cowshed; people who do not have a bed, sometimes no straw, to lay their numbed limbs down to rest, who are lacking everything except distressing poverty, deep misery.” (Perneel, 2000, 249).

After almost a century, in this paper we study again the effect on mortality of the extreme Dutch winters and summers of the nineteenth and twentieth centuries. We focus on those topics which already attracted the attention of medical doctors at the time: Were some age groups more vulnerable than others? Were there social classes that had to endure the effects of heat and cold more than other groups? Were there indeed short-term and longer-lasting effects of heat and cold? In contrast to the

(27)

nineteenth-century studies, we combine detailed mortality and temperature data with advanced time-series methods to shed new light on the effects that heat waves and cold spells had on mortality in the past. Questions such as these have only received limited attention by historians. Medical doctors and public health specialists

had lost their interest in the temperature-mortality link after WWI2_{, for various}

reasons. One is that water- and food-borne and air-borne infectious diseases, the spread of which was more sensitive to respectively extremely high and extremely low temperatures, lost their importance. Deaths among infants and children, the group most vulnerable to high and low temperatures, became rare occurrences. And, very important, economic and technological developments–improved housing, better working conditions, reduced outdoor employment, better transport, to name only a few–reduced exposure to and the negative consequences of extreme temperatures. This all changed when in August 2003 Western Europe experienced an unprecedented hot summer (probably the hottest in Europe since 1500; see Luterbacher, Dietrich, Xoplaki, Grosjean, and Wanner, 2004), with deadly conse-quences for the population. Detailed analyses of the excess mortality related to that heat wave were published for many European countries (for France, see e.g. Rey, Fouillet, Jougla , and Hémon, 2007). Haines et al. (2006) argued that “climatologists now consider it very likely that changes in climate have doubled the risk of a heat wave such as that experienced in 2003”. We will attempt to show here that the relation between extreme weather conditions and mortality and the changes therein over time are an interesting research topic for historians too. How societies coped with environmental shocks such as extremely high temperatures and whether or not they were able to restrict their effects on mortality provides us with a valuable measure of societal development (Galloway, 1994). Studying the differences in vulnerability to this environmental stress by social class, age and sex gives us information on the conditions– food, shelter, clothing–under which these groups lived and on the way these conditions changed over time (Bengtsson, 2004). This is nowhere better illustrated than in Eric Klinenberg’s Heat wave: a social autopsy of

disaster in Chicago (Klinenberg, 2002).

Compared with a previously published study in which we also analyzed the changing temperature-mortality link, the present study encompasses data for four provinces instead of the single one that was the topic of our earlier paper (Ekamper, Van Poppel, Van Duin , and Garssen, 2009). This not only allows us to reach firmer

2_{High death rates due to winter cold remained a topic in temperate zones of Western Europe as} thousands of extra persons died there in extremely cold winters (Analitis et al., 2008; Baccini et al., 2008; Healy, 2003; Keatinge et al., 1997; McMichael et al., 2008).

(28)

19

conclusions, it also makes it possible to compare regions that differ in their microclimatological environment.

The temperature-mortality link as topic for historians

There have been few quantitative assessments of excess mortality during heat waves or cold spells for historical populations, let alone studies in which the vulnerability of specific groups was analyzed. Historical studies of the link between extreme weather conditions and mortality are, as a rule, based on rather crude weather and mortality indicators. Mortality is mostly available only for the population as a whole, without distinction by sex or age, and often only on a monthly basis, whereas temperature data tend to be monthly averages (see e. g. Galloway, 1985; Galloway, 1986; Galloway, 1988; 1994; Landers, 1986; McDowall, 1981). An important drawback of a monthly aggregation of mortality data is that it makes it almost impossible to properly identify the effects of extreme heat or cold on mortality. Deaths caused by extreme heat, for example, appear to occur at very short lags (0-1 days) and it is therefore highly likely that an effect of heat-related mortality will be attenuated even in a weekly-aggregated analysis (Carson, Hajat, Armstrong , and Wilkinson, 2006). Historians rarely have access to information on daily numbers of deaths and daily temperature data, and where that is the case these records are collected only for small communities and restricted time periods.

The number of studies in which the relation between extreme weather conditions and mortality is studied over a long period of time with adequate methods and on the basis of comparable data is extremely limited. Contemporary epidemiological studies rarely cover a period long enough to encompass major economic, demographic or epidemiological transitions (Carson et al., 2006). There is not a single study in which advanced time-series methods have been used to study mortality displacement during heat waves and cold spells for historical populations. Recently, Rau (2007, 13-14) noted that “surprisingly, there is not much literature in the field of seasonal mortality on the ‘classical’ social mortality determinants such as income, deprivation, wealth, marital status, education, occupation”. He also noted that “most of these analyses […] studied the same country (UK) using similar methods based on ecological data”. Rau was referring to the present-day situation, but his conclusion applies even more to the historical study of the effect of extreme weather conditions. Although during heat waves and cold spells contemporaries often referred to the effects that extreme weather had on the poor in particular, there is hardly an empirical study focusing on this aspect.

(29)

In past decades historical databases with more detailed information on deaths have become available for a number of countries. This paper uses data relating to four of the eleven provinces of the Netherlands, covering a period of 100 years and allowing us to study the effect of extreme temperatures separately by age, sex and social class. We relate these mortality data to series of standardized, location-specific daily temperature measurements, and in doing so we make use of sophisticated statistical methods. The data cover the period during which the Netherlands underwent a transition from a mortality regime characterized by high annual fluctuations in mortality due to the dominance of infectious diseases (lasting until around 1875) to a regime in which infectious diseases disappeared almost completely and degenerative diseases became the most important cause of death. This transition of the cause-of-death pattern was accompanied by a changing age profile of death, in which no longer infants but the highest age groups accounted for the majority of deaths.

The period that we study is also an interesting one because it was characterized by strong socioeconomic progress, which might have caused a reduction in the vulnerability of the population to external circumstances: national income grew rapidly after 1860, housing conditions improved, clothing became better, and food and fuel became widely available. As we are able to compare regions with different levels of economic development we have an excellent opportunity to find out how vulnerability to extreme circumstances changed over time and varied by region.

Of course our dataset also has some drawbacks. Periods of extreme heat and extreme cold are scarce in the Netherlands, with maximum temperatures only sporadically exceeding 27°C and minimum temperatures rarely dropping below – 10°C. As a consequence, the variation in temperature-related mortality is small by international standards (Healy, 2003; Keatinge et al., 1997; Keatinge et al., 2000; McKee, 1989). On the other hand, studies have documented that in countries with harsh climatic conditions during winter, winter excess mortality is lower than in countries with relatively warm or moderate climates, and this same mechanism applies to the excess mortality during summer. This “seasonality paradox” (Gemmell, McLoone, Boddy, Dickinson and Watt, 2000), resulting from the fact that the population is not accustomed to protecting itself adequately from uncommon temperatures, might have led to strong effects even in a country with a moderate climate like the Netherlands.

Unfortunately, we have no information on climatic conditions other than temperature which might have an effect on mortality, such as humidity, wind speed or wind direction. Nor do we have information on temperature-related variables such

(30)

21

as air pollution and influenza, which might have played a role in (changes in) weather-related excess mortality.

Mechanisms in weather-related mortality

A large number of studies present overviews of the factors that account for an increase in mortality due to cold or heat in contemporary societies. As Keatinge and Donaldson (2004, 1094-1095) have made clear, few of the excess deaths during cold are due to the body simply cooling until vital organs such as the heart cease to function, and few heat-related deaths are due to hyperthermia, overheating of the body. Cold-related deaths are mainly caused by coronary and cerebral thrombosis and respiratory diseases, whereas the same thromboses account for most heat-related deaths. The precise effects of extreme heat or cold depend not on

temperature as such alone3_{but also on specific conditions in which the temperature}

decline or rise took place and on other climatic conditions.

The effects of cold and heat may consist of a more or less instantaneous effect and a more delayed effect. Temperature falls in winter are closely followed by increased mortality, with characteristic time courses for different causes of death. For heat periods too, immediate effects (such as acute myocardial infarction) as well as long lag times might be distinguished. The length of the period of heat and cold might be a factor determining the effect on mortality. For heat and for cold it might be assumed that the effect on mortality is higher the longer the period during which the temperature is extreme (Huynen, Martens, Schram, Weijenberg , and Kunst, 2001). Main heat effects are usually visible on the current day or may last another day or two (Pattenden, Nikiforov , and Armstrong, 2003). Compensatory effects on

mortality might be registered when longer time periods are studied. The number of

deaths caused by heat waves is often assumed to be compensated for by a fall in number of deaths in subsequent weeks. The suggestion is that heat mainly has an effect on people whose health is already impaired and who would have died within a short time anyway. This compensating effect is known as “harvesting” effect (Huynen et al., 2001). However, no general agreement exists among scientists on the length of the period over which harvesting effects can be expected, which vary from a few days or weeks in the short term to several months or even years in the longer run (Toulemon and Barbieri, 2008). The effects of heat and cold might also be contingent on the sudden occurrence of a change in temperature. Such effects depend on

3_{There is also some debate concerning the comparative impact of minimum, maximum and average} temperatures on mortality (Kalkstein and Davis, 1989).

(31)

whether populations have had the chance to adapt to extreme weather conditions (Ballester, Michelozzi , and Iniguez, 2003). Effects of outdoor air temperature might be modified by other weather conditions, such as high humidity and strong air flow (Gill, Davies, Gill , and Beevers, 1988). A study on daily variation in mortality in relation to temperature and two wind-chill indices for the Netherlands (1979-1987) showed that hazardous weather situations could be identified almost as accurately by temperature as by an index that also included wind-chill (Kunst, Groenhof and Mackenbach, 1994). Effects of high and low temperatures also depend on the

climatological situation of the region studied. Studies of populations living in widely

different climates show that they have adjusted to their own climate remarkably effectively over time. This applies to cold as well as to hot regions (Keatinge et al., 2000). Countries with the mildest winter climates exhibit the highest effect in winter mortality (Healy, 2003). Breschi and Livi-Bacci (1994) and Oris et al. (2004, 392-393) showed that winter peaks in mortality among infants were more common in climates with mild winters than in harsh climates where the population had a high capacity for adaptation: thus in temperate regions winter is a more dangerous and impacting period than summer, although clothes, heating and good housing could reduce its effects.

It is important to stress that the temperature-mortality link might be due to

mechanisms other than the direct effects of exposure of the human body to extreme

temperatures. In particular for historical populations, these indirect effects on mortality cannot be neglected. We mention here two of these mechanisms. Extreme weather has a direct effect on biological processes that are crucial to man’s survival, such as the growth of food plants and animals, and on the physical environment, such as flooding and storms. These direct effects could lead to second-order effects on mortality (Michaelowa, 2001). Relevant is also the link between temperature and the incidence and virulence of infectious diseases, the most important cause of death until the first decades of the twentieth century. Temperature and rainfall affect the mobility and strength of pathogenic micro-organisms and those of insects and animals that carry them. Where sanitation was virtually unknown and water supplies subject to contamination, warm summers promoted the spread of infectious diseases through increased proliferation of animal, insect and bacterial vectors (Galloway, 1994). Cases in point are malaria, diarrhea and other gastric conditions, the latter particularly affecting those children who had lost the protection of the mother’s milk (Oris et al., 2004).

In studies dealing with present-day effects of extreme temperatures on mortality the question often is whether there are specific groups whose health is more affected by extreme heat or cold than others. Usually the focus lies on gender

(32)

23

and/or specific age groups, and physiological factors are used to explain differences. Significant variations in effects of heat and cold according to age have been related to variations in thermoregulatory function and appreciation of cold and heat with age. This is considered the main reason why the elderly are disproportionally affected by extreme weather conditions (Hajat, Kovats, and Lachowycz, 2007). There is no evidence in present-day studies of excess mortality attributable to heat waves in children (Kovats and Kristie, 2006), and only rarely is mention made of the effect of extreme cold on the death risks for this age group. The determination of which gender is more susceptible to weather fluctuations is much in dispute. In studies of England, Wales and France women had higher heat-related mortality, reflecting adverse effects of menopause on thermoregulation (Hajat et al., 2007; Rey et al., 2007). For cold-related mortality, gender differences were not significant (Keatinge et al., 1997).

Relatively little present-day research has examined variation in temperature vulnerability by socioeconomic position, and the few existing studies often present conflicting results. O’Neill et al. (O'Neill, Zanobetti, and Schwartz, 2003) observed stronger cold and heat effects among the less-educated in most of the seven US cities they studied. Such an effect was not found in a Spanish study (Borrell et al., 2006). Naughton et al. (2002) found increased risk of heat-related death during the 1995 Chicago heat wave among low-income residents, whereas Kaiser et al. (2007) found the same effect among the lower educated. McDowall (1981) observed higher winter excess mortality in England during the 1959-1972 period among semi-skilled and unskilled workers than among other social classes. Donaldson and Keatinge (2003) observed for 1998-2000 in England and Wales that cold-related mortality in men of working age was low for unskilled occupations but high among men of retired ages in that same social class. The beneficial effect of work-related factors in this social class was explained by internal heat production from manual work, offering protection against daytime cold stress. Other authors have argued that unacceptable working conditions during high temperature periods can lead to increased mortality in lower social classes. Rau (2007, 127-162) studied individual-level data for Denmark for 1980-1998, using a variety of socioeconomic indicators. He did not observe a connection between excess winter mortality among people aged 65 years or older and factors such as educational level, wealth and housing conditions. Area studies give conflicting results too. A study for the 1993-2003 period in the UK (Hajat et al., 2007) observed very little difference in heat effects according to level of deprivation of the neighborhood and no link between cold and deprivation. Results of a study of the French 2003 heat wave however point to the most deprived populations being more vulnerable to heat waves (Rey et al., 2009). It remains to be seen whether in

(33)

the nineteenth and early twentieth centuries such a difference in vulnerability also can be observed.

Setting, data and methods Study regions

Nationwide and compulsory birth and death registration according to the rules laid out in the Napoleonic Code was introduced in the Netherlands in 1811, at the time of incorporation of the Netherlands into the French Empire. In recent decades, dozens of staff and volunteers in Dutch provincial archives have started to enter death records into a database within the framework of projects called ISIS and GENLIAS. The purpose of these projects is to build a database with genealogical information on all marriages, deaths and births taking place in the Netherlands from the introduction of the vital registration system (1811) up to when such data were not yet in the public domain. Death records enter the public domain after 50 years. We were able to use data for four of the eleven provinces of the Netherlands: Drenthe,

Gelderland, Groningen, and Zeeland.4_{These provinces were selected because entry}

of death certificates has been completed for the whole of the province and because the information entered in the database includes information on sex, age and

occupation of the deceased.5_{Figure 2.1 gives an overview of the location of the}

selected provinces.

The four provinces each have their own particular ecological, social and economic structure. Gelderland is located in the central eastern part of the country, extending from the German border westward to the former Zuyder Zee. In the northwest the hill plateau of the Veluwe was a wasteland covered with heath and some woods, ill-adapted for cultivation and of little economic value except for some wood-cutting and paper mills. The fertile marshy area of the Betuwe between the Rhine and the Waal supported orchards, market gardening and mixed farming. The southwestern section was a long, narrow westward extension along the Rhine river with brickyards and dairy farming. Some textile works were located to the east. Small regional marketplaces and several larger towns such as Arnhem and Nijmegen hosted industrial activities and administrative services. Farms in Gelderland were

4_{Sincere thanks go to the Drents Archief (Archives of the province of Drenthe), the Gelders Archief} (Archives of the province of Gelderland), the Groninger Archieven (Archives of the province of Groningen), and the Zeeuws Archief (Archives of the province of Zeeland) for making their data available.

5_{Data for Gelderland covered only municipalities alphabetically up to the letter V, and relate to 75 per} cent of the total number of deaths in the province.

Time and space in historical demography: some case studies using Dutch micro-data

University of Groningen

Time and space in historical demography

Ekamper, Peter

Time and space in historical

demography

Time and space in historical

demography

Some case studies using Dutch micro-data

PhD thesis

Peter Ekamper

Supervisors

Assessment committee

Table of contents

1.

Introduction

2.

Heatwaves and cold spells and their

effect on mortality

An analysis of micro-data for the Netherlands in

the nineteenth and twentieth centuries