Historical demography strand abstracts


Historical demography of Africa & the Pacific. Monday 12 September 16:45pm   

Patterns of morbidity and mortality in colonial Uganda
Shane Doyle, University of Leeds 

This paper examines the factors which shaped morbidity and mortality patterns in colonial Africa. Census data, vital registration and annual medical reports indicate that mortality fell substantially over the course of the colonial period, most rapidly amongst infants, and seemingly due primarily to a decline in epidemic disease. We have, however, little sense of how disease patterns changed after the decline in epidemic mortality between the wars. Knowledge of how disease incidence varied by factors such as ethnicity, religion or educational attainment is also very limited. One means of enriching our understanding of who sought medical attention, and how disease outcomes varied according to the characteristics of the patient, is to analyse hospital records. These are underused resources in African historical demography, in part because so few hospitals retained patient registers. This paper discusses the trends observable in registers from several Ugandan mission hospitals, examining how the distribution of disease, length of hospital stay, and the probability of being cured varied by factors such as religious belief. It also discusses the uneven decline in the incidence of particular diseases over time. The results show significant variation by religion and education for some but not all communicable diseases, lending some support to the scholarship which emphasises the transformative behavioural impact of conversion and schooling, but also indicating that public health campaigns benefited the entire population. The discussion of the data is supported by regression analysis. 


Censoring and compositional effects in African parish registers
Sarah Walters, London school of Hygiene and Tropical Medicine 

Using parish registers to reconstruct Africa's demographic pasts raises issues of censoring and selection. I use examples from the Counting Souls Project to describe the problem, and to suggest solutions. The project includes data on over half a million events from nine of the oldest Catholic parishes in Malawi, Tanzania, Uganda and Zambia. The parishes were founded by missionaries in the late-nineteenth century and data are available until 2012. European historical demography has relied on a long-established Church, with parish registers often spanning several centuries. In contrast, the East African parishes span a single century and include the formative years of the Church, when congregations were unstable and religious identities were fluid. First generation Christians tended to be adults, implying a huge degree of left-censoring which must be taken into account in analyses. In the absence of alternative secular sources of longitudinal microdata, the registers are used until today and this, together with the poverty of death recording, implies incomplete life histories and right-censoring. Further censoring and compositional effects occurred following political Independence and doctrinal and administrative shifts. I outline the biases that would arise by ignoring these effects. Solutions include using time-to-event analysis and denoting observation time in novel ways, as well as exploiting fully any relational data. A deeper qualitative understanding of the social history of each parish is essential to reveal the more subtle censoring and compositional effects of a growing church and modernising population. 


The number of native Hawaiians and part-Hawaiians in Hawaiʻi, 1778 to 1900: Demographic estimates by age, with discussion
David A. Swanson, University of California Riverside 

A high level of uncertainty surrounds the size of the Hawaiian population at the time of first European contact in 1778. Estimates range from 200,000 to 1,000,000. While some estimates have more of an empirical base than others, none of them takes advantage of the high level of momentum found in demographic processes, something that is done in this paper using “backcasting,” a demographic forecasting method run in reverse from known data. Using a commonly used technique for this purpose, the 1910 count of Native Hawaiians by age in Hawaiʻi is taken back to 1770 in decennial cycles. Interpolating between the 1780 and 1770 estimates yields an estimated 683,200 Hawaiians in 1778. In addition, an estimate is made for the number of Part-Hawaiians in Hawaiʻi and decennial estimates from 1780 to 1900. Comparisons are made in terms of the historical demographic dynamics of the Native and Part-Hawaiian populations along with some references to their current demographic situations. Suggestions are made for future research on these two populations. 


Social Mobility among Christian Africans: Evidence from Ugandan Marriage Registers, 1895-2011
Felix Meier zu Selhausen 1, 2, Marco H.D. van Leeuwen 3, Jacob Weisdorf 1, 1 University of Southern Denmark, 2 University of Sussex, 3 Utrecht University   

Did the colonial era prevent or provide African social mobility in Africa? We use hitherto unexplored Christian marriage registers from colonial and post-colonial Uganda to investigate long-run trends in rural and urban African inter-generational social mobility. We find that the colonial era disrupted pre-colonial social classes and created a period of enhanced mobility and more equal opportunity. In urban areas, the colonial labour market was the main ladder for upward mobility, and the mission society provided the basic education and social reference needed to climb it. In rural areas the absence of a broad colonial labour market made social mobility less common and mostly linked to employment in mission schools and hospitals. Aptitude gradually ascription with sons of the traditional landed aristocracy (chiefs) gradually losing their social advantage through the creation of a new commercial and technical class, affording opportunities for mobility. This continued throughout the post-colonial period despite extended episodes of political turmoil. Our data reject the ‘buffer zone’ hypothesis that sons of blue-collar fathers were prevented from entering into white-collar work. 


Population & institutions. Tuesday 13 September 09:00am 

Rethinking Adam Smith on Population
Philip Kreager, Oxford University 

Adam Smith’s pre-eminence in the foundation of modern economics has in general not been shared by demographers, who give vastly more attention to Malthus. At first, the reasons for this seem evident: Malthus put rapid population growth at the centre of his account of modern economy and society, whereas Smith considered problems of population growth as resolved normally – before any crisis could emerge -- by the division of labour, and this was the case whether that division was managed fairly or badly. Hence Smith’s comments on population appear as passing adjuncts of other matters, not a core component of his theory. One might add to this the difficulties posed by the sheer scale and subtlety of Smith’s oeuvre: the Wealth of Nations runs to nearly a thousand pages, and is not a stand-alone text, depending closely on his Theory of Moral Sentiments, many essays, and lectures. Recent historical research on the structure of population thought before the 19th-century rise of statistics sheds new light on Smith, and makes his reasoning on population much more accessible. It turns out that the mechanisms that tend to slow economic and population growth are central to his political economy and, indeed, his whole view of history. Contrary to the currently popular view of Smith as advocate of a benign invisible hand in economic affairs, he saw such slowing as by nature endemic. This view is very applicable to understanding current low levels of fertility in Europe, and the heterogeneity of fertility declines more generally. 


Demography and democracy: Age structure, political participation, political competition and democracy in Europe, 1810 – 2000
Moshi Optat Herman, Mike Murphy, The London School of Economics and Political Science 

Even though it has long been assumed that demographic structure may affect political outcomes, research is limited. This paper examines the effect of age structure measured by mean age and three indicators of economic dependency (proportion young, proportion working-age, and proportion elderly) on the quality of political institutions in Europe from early 19th C to the present. We draw from Europe’s unique context of a progressively aging population and its political dynamism in the last two centuries. We examine eight countries with complete data including Denmark, Finland, France, Iceland, Italy, Netherlands, Norway, and Sweden. Data and methods: For demographic variables, we use data from the Human Mortality Database (HMD). We use two sources of data for political variables. We use the Polyarchy dataset to devise three measures of quality of political institutions including the extent to which there is political competition in a country, the level of political participation, and overall level of democracy (a weighted average of political competition and participation). In addition, we use the Polity IV dataset as alternate measures of democracy which focus on availability of civil and political liberties and institutional constraints to political power. We apply panel data methods initially using Fixed Effects models to examine associations between changes in age structure (mean age and proportions young, working-age, and elderly) on political competition, participation, and democracy. We control for other demographic characteristics (total population, and adjusted total fertility rates), aggregate economic output, and socioeconomic characteristics (educational attainment, economic inequality). 


Geographical variations in the proportion of old people in Victorian England and Wales
Tom Heritage 1, David Clifford 1, Andrew Hinde 2, 1 Department of Social Statistics and Demography, University of Southampton, 2 Southampton Statistical Sciences Research Institute    

This paper examines geographical variations in the proportions of males and females aged 60 years and over in England and Wales, using data from the 1861 census. Variations in the proportions of old people in any area can be understood as a consequence of past fertility and mortality rates and migration patterns. The latter are related to the economic and occupational structure of each region. We estimate models to quantify the association between the proportion aged 60 years and over and occupational structure, controlling for past fertility and mortality (which have the expected effects on the age structure). Old people were more numerous where a high proportion of adults had income from property, but some other occupational effects (e.g. of mining and manufacturing) are small after controlling for fertility and mortality. We then examine the geographical distribution of the residuals from these models to identify areas with unusually large or small proportions of old people. The paper then examines the association between the proportion of old people in a given area and their treatment by the Poor Law. We consider the argument that, in areas where old people were relatively numerous, they were viewed by Boards of Guardians as a distinct category of the poor requiring specific relief policies and practice, whereas in areas where there were few old people, they were regarded as individuals, which normally meant that they were treated less generously. 


Historical demography of Britain from micro-census data 1851-1911. Tuesday 13 September 13:30am 

The Geography of the Fertility Decline in England and Wales
Alice Reid 1, Eilidh Garrett 2, Hannaliis Jaadla 1, Joe Day 1, 1 University of Cambridge, 
 2 University of Essex   

The first demographic transition, during which nations experienced major falls in both fertility and mortality, gathered momentum in Europe during the last quarter of the nineteenth century. Despite extensive research, the process of fertility decline - how and why it happened, where it started, and who led the behavioural changes involved - is still not well understood. Our lack of understanding of fertility decline has, in part, stemmed from a widespread lack of the data necessary to measure age-specific fertility at either a fine geographical scale or on a detailed social or occupational basis over the course of the transition from high to low fertility. Age-specific measures are essential if insight into ‘starting’, ‘spacing’ or ‘stopping’ fertility regulating behaviour is sought. A time series of such measures across geographical and social space is also vital when trying to identify how new forms of behaviour spread through the population. This paper will use the own-child method to calculate age-specific and total marital fertility rates for small geographic areas in England and Wales and for social classes within those areas in 1851, 1861, 1881, 1891, 1901 and 1911. The paper will present the first detailed spatial and social pictures of age-specific fertility decline over late nineteenth-century England and Wales, and will offer a preliminary multivariate spatial statistical analysis of the patterns. 


Marriage patterns in Victorian England and Wales: 1851-1911
Eilidh Garrett 1, Kevin Schürer 2, Joe Day 3, Alice Reid 3, 1 University of Essex 2 University of Leicester, 3 University of Cambridge   

When Bob Woods (2000, Figure 3.9) wished to illustrate the relationship between three main measures of nuptiality: the percentage of women never married, female SMAM and Im, he had to use Knodel’s data for Germany in 1880 because, he lamented, such information was not available for English and Welsh registration districts. With the release of the Integrated Census Microdata (I-CeM) dataset, which includes all individuals enumerated in each census from 1851-1911 (bar 1871) it is now possible to calculate marriage profiles over space, through time and across occupational groups for England and Wales. This paper will report on work using the ICeM data to assemble measures of nuptiality at the sub-district level for both males and females across the 1851-1911 period. It will then move on to use the information on birthplace data given for each individual in every census to compare measures of nuptiality among those born in selected communities, wherever they were living within England and Wales on a particular census night, with the nuptiality of residents within the chosen communities at the same date. It thus becomes possible to compare the nuptiality of ‘stayers’, ‘in-migrants’ and ‘out-migrants’, providing a much more detailed understanding of the inter-relationship between migration and nuptiality and how the former affects measurements of the latter; a longstanding puzzle for historical demographers. 


The Geography of Early Childhood Mortality in England and Wales, 1851-1911
Hannaliis Jaadla, Alice Reid, The Cambridge Group for the History of Population and Social Structure, University of Cambridge   

Using published mortality statistics from Quarterly, Annual and Decennial Reports of the Registrar General and the individual-level census data from the Integrated Census Microdata (I-CeM) project, this study will illustrate spatial and temporal variations in infant and early childhood mortality from 1851 until 1911. It is well known that there is considerable spatial variation in infant and early childhood mortality and that the trajectories of decline in early age mortality both displayed different trends and were differentiated by place. Up to now, the most detailed spatial analyses of nineteenth century mortality on a large scale used decennial data for the 614 registration districts in England and Wales (Woods and Shelton 1997). However, studies of selected smaller areas have shown that even registration districts can contain considerable variation in the risk of death due to variations in population density, occupational and industrial make-up, environmental hazards and local-disease environments (Garrett et al 2001, Garrett and Reid 1994, 1995, Reid 1997). Therefore, this paper will use newly generated data for infant mortality for the over 2,000 registration sub-districts of England and Wales to examine changes in infant and early childhood mortality during the second half of the nineteenth century. It will also explore differences in the relationship between infant and early childhood mortality over space and time and consider different methods for estimating early childhood mortality (age 1-4) for sub-registration districts.


The Importance of London, 1851-1911 Using the Individual-level census data to reconstruct lifetime migration paths
Kevin Schürer, University of Leicester, Joseph Day, University of Cambridge   

Almost all studies of historical internal migration have looked at migrants as a proportion of the population in which they are resident. This paper will turn the telescope around and will produce the first systematic analyses of migration using individuals’ place of birth as reported in the 1851-1911 census enumerators’ books (CEBs). Using individuals’ recorded place of birth it is possible to analyse the determinants of migration at a far greater degree of spatial resolution. Indeed, by linking individuals’ place of birth to a geographical information system (GIS), it becomes possible to analyse how and why migration streams shifted, by relating these changes to the socio-economic context of both individuals’ parish of origin and residence. However, this paper will focus on the pattern and determinants of in-migration to London from 1851-1911. This seems an appropriate starting point for an analysis of migration between 1851 and 1911 given that London has long been an important destination for migrants. Through an analysis of the changing migration streams into London, the broader determinants of migration throughout England and Wales can be better understood. It is also possible to use census data to model the determinants of London-bound migration, and identify whether, for example, London attracted more migrants than might be predicted from distance and economic opportunities alone. By analysing the migration process not just in one census year, but in multiple census years from 1851-1911, it can also be determined whether the importance of London as a destination changed over the period.


Health & living conditions in the European past. Wednesday 14 September 09:00am 

Urban sanitation and infant mortality. Sewerage coverage and health improvement in the city of Madrid in the early twentieth century
Diego Ramiro-Fariñas, Yolanda Casado-Ruiz. CCHS-CSIC. Centre for Humanities and Social Sciences, Spanish Council for Scientific Research   

Since the end of the nineteenth century, improvements in urban infrastructures started to be implemented in many European capitals, with the main objective of improving health of their inhabitants. In the case of the city of Madrid, due to the high mortality rates, poor hygiene and sanitary conditions were an important concern. As in other urban areas, one of the main measures to tackle high mortality was the improvement of water and sewerage systems. Therefore this paper will focus on how the improvement and extension of the urban sewerage systems in the city of Madrid made an influence on the decline of infant mortality, and infant mortality by specific causes of death (Diarrhea, enteritis and typhoid fever) and in what way did it happen. To do this, we will use individual records of deaths in the city of Madrid in two different periods of the early twentieth century; 1905-1907 and 1920-1922. This will allow us to know the place of residence and place of death of the inhabitants and compare with data on sewerage systems expansion and explore in which way this expansion affected child health in a big urban environment. 


Poverty, malnutrition and disease: tuberculosis in Sardinia between the end of the 19th Century and the Second World War
Lucia Pozzi, University of Sassari   

There is a vast literature on the presence of tuberculosis in the past Italian demographic experience. However, the South and the rural areas of the country have seldom been explored. This study analyses the demographic burden of disease in Sardinia between the late nineteenth century and the Second World War, focusing not only on the island's main towns, but also the predominantly rural areas strongly affected by the presence of tuberculosis. The reasons for the choice of Sardinia are many and of different origin. First, Sardinia was the only Italian region to know a sharp increase in tuberculosis mortality in the period examined to the point that in the 1920’s and 1930’s the Sardinian provinces were amongst the most affected in the Italian Kingdom. Moreover, in the case of Sardinia, cause specific mortality statistics allow to document all major phases of the experience of the disease, including the epidemic stage and its peak. The experience of TB, a multifactorial disease with a strong social connotation, also provides the opportunity to document the conditions of poverty, hunger /malnutrition, and overcrowding that characterised the vast majority of the Sardinian population, urban and rural. The study also deepens our understanding of age and sex tuberculosis mortality differentials, the relations between TB and other poverty-related diseases, such as malaria and trachoma, encouraging a better understanding of the living conditions of the Sardinian population.


Does occupation matter? Child mortality and fertility in eight English parishes, c. 1790 to c. 1850
Ellen Potter, University of Cambridge   

Today, health differences generally mean health inequalities. In England, as elsewhere, affluence is strongly correlated with good health and low mortality. It is not, however, clear that this has always been true. This paper investigates mortality and fertility differences in England during the early nineteenth century, asking (1) whether differences were consistent across settlement types, and (2) whether differences had already become inequalities. The data consists of eight reconstituted parishes, covering the period c. 1790 – c. 1850. Families are categorised into social groups using measures of economic, cultural, and social capital. These are estimated using: the capital intensity of the husband’s occupation; the literacy of both husband and wife; and HISCAM, which scores the husband’s occupation based on social distance. Differences in infant and child mortality are assessed through survival analysis. Due to the less complete fertility data available, differences in fertility are assessed through birth analysis and, where complete reproductive histories are present, children ever born. Preliminary results suggest that in early nineteenth century England, mortality and fertility did differ by social group. However, these differences were not consistent across settlement types. Nor did the advantage consistently lie with the most privileged groups, suggesting that health differences had not yet become health inequalities. 


Cholera epidemics as a ‘sanitary test’ of British towns, 1832-1866
Romola Davenport, University of Cambridge 

The malign contribution of northern industrial cities to the stagnation of national life expectancy over the period 1820-1870 forms part of one of the most long-running debates in English economic history, regarding the impact of early industrialisation on living standards. The pessimistic view argues that industrial cities experienced a worsening of mortality especially in the second quarter of the nineteenth century, and that this was due to the peculiar conditions of industrialisation, including administrative breakdown and rising social inequalities. The period 1820 – 1850 is particularly difficult to study because of a paucity of demographic data especially for towns. Here we use the official reports occasioned by the cholera epidemics of 1832, 1849, 1854 and 1866 to investigate sanitary conditions in British towns through the lens of water provision and sewage disposal. We argue from the chronology and geographical patterning of the epidemics that cholera spread rapidly through the population and that reported cases and deaths were only the visible peaks of a very widespread and high incidence of infection. Under these circumstances the key determinant of mortality was the extent to which infected sewage contaminated drinking water. The geography of reported cases and deaths did not indicate that northern towns were especially deficient in this respect. Instead cholera mortality was associated with access to rivers and with mining, and precocious development of sewage systems may have aggravated epidemics. We discuss the consequences of our findings for mortality from other waterborne diseases in this period, and their relationship to diarrhoeal mortality in young children. 


‘The English Disease’: rickets after 1600 – diagnosis, environment and social customs of childcare
Gill Newton, University of Cambridge   

There have been two major panics in England over seemingly lethal epidemics of rickets in infants, besides concern over the lifelong skeletal deformities and weakness it could inflict upon survivors. The most recent and best-known was that of the late nineteenth century, which led eventually to scurvy being separately medically identified and understood as the culprit. The other took place during the turbulent mid-seventeenth century, when physicians’ interest in a new disease that was apparently killing ever-growing numbers of children, especially of the wealthy urban elite, was initially slow to develop. Apothecaries and preachers delivered the first written mentions of rickets, a name bestowed by the women Searchers who first recognized it. Rickets may be caused by a lack of Vitamin D, obtained mainly from exposure to sunlight; but also by insufficient calcium, which is found in specific foods including milk. Infant feeding practices during both panics were in a state of flux, with exclusive breast-feeding uncommon among the wealthy. The mid-seventeenth century rise in rickets deaths followed a surge in mineral coal use, then an unprocessed, smoky fuel that commentators were observing induced sunlight-blocking smog, and on which British cities were already uniquely dependent. However, recorded rickets deaths subsequently dwindled to near zero by the early eighteenth century. This paper will investigate how rickets first rose to prominence in England from 1634 onwards, who and where was affected, what was understood by the disease, and possible reasons for its cessation as a cause of death while continuing to be prevalent. Evidence is drawn from Bills of Mortality, parish registers, osteological studies, and references in lay and medical writings.