Surviving the loss of the parent in a nineteenth-century Dutch provincial town.
Until late in the nineteenth century, the loss of one or both parents was an event with which many children were confronted. As a consequence of the low life expectancy, the high age at which reproduction started and the large number of children within marriage a substantial proportion of children had lost one or both of their parents by the time at which they reached adulthood. Data for the Netherlands for the period 1850-1900 show that between 8 and 11 percent of all persons aged 20 or less had lost one of their parents and another 1-2 percent had lost both parents.(1)
The loss of a parent usually endangered the smooth operation of the family and could produce dramatic declines in the economic and physical well-being of the surviving spouse and children.(2) Generally speaking, the negative consequences of orphanhood for the child would be greater in the case of the loss of both parents. The child would fare worse materially and psychologically and there would be more need for community intervention than in the case of incomplete orphanhood. When there was no surviving parent to continue the parental role the child who had lost both parents had to be adopted, put into service, institutionalized, or by default launched into premature independence, depending on his or her age and other personal characteristics, the prevailing culture and social structure, and other life contingencies such as numbers and location of kin and the state of the economy.
Strangely enough, we do not know very well how orphans in the past fared materially and psychologically and whether or not the problems associated with orphanhood were greater in case of the loss of the biological father, the biological mother or both parents. Given prevailing systems of sex stratification and division of labour, one might assume that a widower was usually better able to provide economically for his orphaned children and more able to purchase childcare services which he himself could not provide. Yet we do not know whether this could fully compensate for the loss of the care that the mother usually provided. Only a general impression from diaries, letters, autobiographies and similar sources can be given. The same applies when it comes to the consequences for the child in case the surviving spouses tried to solve the problems with which they were confronted by remarrying. Entry into stepchild status in a sense normalized one's family situation, but the effect on the life chances of the child, at least as depicted in fairy tales, was not always happy. On these aspects too, the surviving empirical evidence is virtually silent.(3)
In this paper, we will pay attention to one of the possible consequences of orphanhood for the life of the child. For this purpose, we take the words life chances literally and study the effects of orphanhood on the survival prospects of children. We study several categories of children - paternal orphans, maternal orphans, stepchildren and children whose parents are still alive - and analyse whether they had different risks of dying. Our study of the effect of parental loss on the situation of the child relates to Woerden, a small nineteenth-century Dutch town.
The first part of the paper gives a brief overview of the mechanisms through which loss of one of the parents might influence mortality risks of the children. It is based on the contemporary literature on the consequences of bereavement for mortality as well as on information on family structure and family functioning in historical populations. We then review the literature on the survival of orphans in a historical context. Next, we briefly describe the historical situation of the orphaned child in the Netherlands, and sketch the socioeconomic situation of our research area. After introducing our database we study the effect of the loss of the parent(s) on the life chances of the child, taking into account other factors which might have had an influence on the survival chances of the child and making use of hazard analysis.
Explaining effects of parental loss on mortality
Two general explanations are suggested for the poorer health and higher mortality which in contemporary populations is often observed among children from maritally disrupted families(4): the association might partly be due to errors in data or methods and thus is a statistical artifact or the association between parental loss and mortality of children is genuine.
Two kinds of artifacts can be envisaged which may cause a relation between parental loss and child mortality(5):
* A common risky event shared by one of the parents and one or more children. Events such as the delivery, a vehicle accident, fires or common drowning, and diseases passed in utero from an infected mother to her fetus, or infectious diseases(6) might all lead to the death of the child during the same day as the parent, or to a child's death within a delay of days or weeks, incorrectly creating an impression of an effect of orphanhood.
* A common unfavourable environment shared by parents and children. Parents and children share many aspects of their daily life: mother and child would most likely eat the same diet, share the same house and breathe the same air. A part of this 'common environment' may be hazardous to health and in such cases both parents and children run a risk of early death.
Parental loss might also have a real effect on the survival of the children which are left behind. In searching for the pathways via which parental loss may influence the survival of the children, we can fruitfully make use of an explanatory scheme developed for the study of the bereavement-mortality relationship after spousal loss. Stroebe and Stroebe have argued that the bereavement-mortality relationship may be generalized beyond spousal loss as the death of any loved person may have fatal consequences for those left behind.(7) Two mechanisms may play a role in the mortality of children after the loss of one of their parents: mortality may be due to indirect changes associated with the status of becoming an orphan or it may be caused by direct psychological factors.(8)
Theories that concentrate on the direct consequences of the loss of the beloved person explain the bereavement-mortality relationship as the effect of grief for the loss of a loved person. Hopelessness, loss of the will to live, and the desolation of grief are seen as mediators in this process. Although the loss of a parent may cause a sense of insecurity, helplessness and hopelessness in very young children, and may bring about frequent crying, sleep problems and somatic symptoms, most authors have the opinion that it is the social circumstances surrounding the loss that bring disruption and disturbance rather than the loss itself. Theories that concentrate on these indirect consequences of bereavement focus on concurrent changes in the lives of the bereaved persons that might lead to heightened vulnerability of the surviving child. The two major theories of this type are stress theory and role theory. In stress models the focus is on situational demands characteristic of widowhood and on the coping resources needed to deal with these demands. Stressful life events play an important role in the etiology of various somatic and psychiatric disorders. They may precipitate the onset of physical or mental disease, directly (via the impairment of the immune function and endocrine change) or indirectly (stressful life events resulting in health-impairing behaviour patterns). Role theories focus on the roles played by persons in different marital statuses that may differentially expose them to risk.
In formulating hypotheses on the effect of parental loss on the child's survival we concentrate on elements from role theories and stress theories. We argue that children are at higher risks of death in situations in which the death of one of the parents is experienced as a more stressful life event by the surviving parent and/or the child and in which the surviving parent and child undergo changes in roles which are harder to adjust to. It is the availability of personal and social resources and how these resources are used that determine whether people are able to adapt to the new situation.(9) The actual distress experienced by children when separated from their parents thus can be moderated if adequate substitute care is provided.(10)
Parental loss in 19th century society
In nineteenth-century Dutch society, women had almost complete responsibility for child care whereas men were mainly employed in non-domestic work. Women performed most housekeeping tasks, being available as primary caregiver, dealing with the affective life of the family, and having closest interpersonal ties with their children. Several factors that could directly influence the risk of morbidity and mortality of children were therefore affected when the mother died.(11) Even setting aside the special situation of the newly-born, the death of the mother could lead to an increase of the level of environmental contamination, to higher nutritional deficiency, to greater risks of injuries and accidents and to less personal illness control (personal preventive measures, quality of care and medical treatment). Historical data are hardly available but contemporary data show for example that when the mother dies children under 12 are more likely than children in fatherless families to experience a negative change in care giving when they are sick, have more responsibility for household chores, have their meals served at a less regular time(12) and often find themselves in unkempt homes. In the recent past, stereotypes about specific gender-roles may have made changes in role functioning after the death of the spouse generally more difficult for men than for women.(13) In addition to that, coping with both their children's and their own grief may have been easier for surviving mothers than for fathers.(14) The mothers tended to do the personal and emotional work of maintaining family connections with extended kin and with friends and as surviving parent they were therefore better able to acquire support from family and friends. Widows and their children were considered as poor, helpless and vulnerable persons eligible for religious admonishments and legal provisions.(15) Especially when very young children were present in the house, an important consequence of the mother's death was that no-one was available to breast-feed the child or that the child had to be weaned prematurely. Continuity in the child's daily life therefore remained more often after the death of the father than after the death of the mother and one might expect that the loss of the mother therefore had a stronger disorganisational impact on the functioning of the family and on the health of the child than the loss of the male breadwinner.(16)
This is not to say that the loss of the husband could not result in serious problems for the surviving widow and her children. Families were strongly dependent on the household head as the main provider of services and goods and that made the risk of falling below the level of consumption where health is affected adversely greater then today. The death of the father could be accompanied by a forced move to lower-standard housing. The father's death could also have indirect negative consequences as part of the mother's time now was required for economically productive activities, and less time could be spent on maternal activities that were conducive to health of the children (cleaning, food preparation). Employment away from home might result in child neglect or care by less skilled older children as infants had to be left alone for long periods of time. Outside work might also have indirect effects on the child's health through a deterioration in maternal health because of long work hours or deficient conditions at the workplace. Nonetheless, we assume that men's economic contribution as fathers could more easily be replaced than women's tasks.
Especially when a widower was left with very young children the surviving spouse was in need of a nurse and a housekeeper for his children. The entrance of a stepfather could also neutralize the economic deprivation of the female-headed household. Entry into stepchild status could normalize the child's family situation. However, the outcome was not always positive. In contemporary United States and Europa, stepchildren appear to be more vulnerable to abuse and sexual molestation.(17) Selective neglect (medical, nutritional, physical or emotional) in comparison to children of complete families in similar socioeconomic circumstances might place stepchildren at greater risk. As we hardly know whether this was also the case in historical populations, we will study whether the life chances of children increased or decreased after remarriage took place in comparison with children living in complete families.
Age of the children could be a key factor in the child's survival prospects after the death of the parent. In the first weeks after delivery maternal death could lead to infant death within hours or days, due to the lack of adequate feeding of the child. Particularly if the resources were few, pathogens abundant and sanitation a luxury, finding substitute food for the new-born was practically impossible. Rochester found that in Baltimore in 1915 the infant mortality rate among babies whose mothers died within two months of childbirth was 625 per 1000.(18) Yet the strong age-dependency is not only related to the availability of breast-feeding. During the first half year of the life of the child, the time the mother spent in food preparation, washing clothes, bathing the child, house cleaning and sickness care had a direct influence on the survival of her child. It is thus in particular the death of the mother directly after the birth of the child which is assumed to have strong effects on the survival of the child.
Sex of the child might also play a role in particular at higher ages of the child. To be sure, recent data from the Harvard Child Bereavement Study relating to children between the ages 6 and 17 did not show significant differences between the direct physical responses of boys and girls upon hearing the news of the death of the father or afterwards. Yet, the situation might have been different as far as indirect effects in historical populations are concerned.(19) Age and sex of the orphaned child might for example influence the way social support is mobilized, and the roles that are played in the family: girls were more prepared than boys to take over cooking and cleaning, and already at low ages they thus could better take care of themselves.
While the death of a parent may have put a child at risk, this risk was not necessarily sustained over time.(20) Most researchers document a gradual improvement in feelings of well-being of the surviving person several months after the loss of the spouse or parent and conclude that the majority of bereaved persons adapt successfully to the new circumstances and learn their new roles to play after the first year or two.(21) Yet it is likely that changes in economic circumstances, such as financial concerns or shifts in the social network, never did return to baseline levels and might continue to fluctuate and require adjustment for many years. The available literature does not give us a precise idea of what effects duration since parental loss may have; therefore, in our study we only can study these effects in an exploratory way.
Historical studies on mortality of orphans
Empirical historical studies on the effect of parental loss on children are still scarce. Only two decades ago Uhlenberg concluded that the experience of losing intimate family members was neglected in historical studies of the family.(22) He related this to the methodological difficulties involved in measuring accurately the effects of mortality and to the lack of the necessary data. Since then, however, progress has been made.
Preston and Haines, using census data for the USA from the beginning of the twentieth century, concluded that children in female-headed households experienced elevated mortality rates.(23) The effect of the husband's absence varied according to working status of the mother. Among non-working women, those whose husband was absent had consistently higher child mortality than those married women whose husband was present. For working women, the situation was usually different. Preston and Haines suggested that the relatively low child mortality of women in this group might reflect their better ability, when the husband was absent, to direct resources towards purchases associated with child survival, even though there may have been fewer resources.
Bengtsson studied the importance of the presence of the parents for the survival of children by using data for the Swedish city of Linkoping in the periods 1797-1810, 1840-49 and 1870-75.(24) While infant mortality was higher for children who lost at least one parent before their first birthday compared to those whose parents survived, parental deprivation did not seem to affect children above the age of one. Of the children who became motherless during their first year of life 60 percent died before age 15, compared with 30 percent of the fatherless and 25 percent of the children with surviving parents. These results contradict those of Hogberg and Brostrom who found in their study on seven mid-nineteenth-century Swedish parishes that the death of a mother during her infant's first year reduced the child's probability of surviving that year from 0.97 to 0.50 and the probability of surviving the next four years from 0.94 to 0.02.(25)
The survival of children born to a mother who died in childbirth has been studied in several other Swedish parishes during the 19th century. Of those who lost their mother at birth only 1.65 percent survived until five years of age. Among children aged between one to five only 13 percent survived until five years of age. However, children who lost their mother after they turned five years of age were almost out of risk and 94 percent survived.(26) A more refined analysis of the survival of orphans, taking differences in the social background of the orphans into account, was done for three parishes from the Sundsvall district in the period 1800-1895. Children who lost their mother or father or both, as infants, between the ages 1-4, or between the ages 5-9 were compared with children still having their parents alive at 10 years of age. Infants losing their mother or father were exposed to a most deadly threat. Only forty out of one hundred motherless infants and seventy out of 100 fatherless infants survived their 15th birthday. Being an orphan by one or two years of age was a very critical stage with a high death risk while above the age of five it seems as if many more children were capable of surviving. 37 percent of the motherless children received a stepmother. Of those infants getting a stepmother 15 percent died before their 15th birthday (Relative Risk = 0.59) while infants having no stepmother run an enormous risk of dying (60 percent: RR = 2.7). Only 7.7 percent of the fatherless group eventually got a stepfather. Among fatherless infants having a stepfather none died while as many as 36.4 percent died among those having no stepfather (RR = 1.75).
Persson and Oberg, using data on foster children in the Stockholm districts of Sodermalm and the Old Town in the 1890s, concluded that compared with children in complete families, children placed with foster parents, had an exceptionally high rate of mortality. In the first year of life mortality was more than twice as high among foster-children (281 per 1000) compared to other infants (109 per 1000). This excess mortality decreased for higher age groups (1-7 years: 24 and 15 per 1000; 7-10 years: 6 and 6 per 1000). As these children mostly lived in households which belonged to the working classes the authors had difficulty in ascertaining in which way this has influenced the results.(27)
More detailed information on the way the household situation of the orphaned child - the age and sex of current household residents, and the nature of their relationships to each other - affected their mortality was collected by Campbell and Lee, in studying data for a village in northeast China during the eighteenth and nineteenth centuries.(28)
Among children aged 2-15 sui, male mortality seems to have been more sensitive to changes in external conditions (price fluctuations, pressure on household resources etc.) than female mortality. There were gender differences in the effect of presence of specific kin, especially parents. But, girls were affected much more by being orphaned than boys. Boys were largely unaffected by whether or not their parents were alive. The implication is that whereas boys could receive care from individuals other than their parents, most likely other members of the household, girls were completely dependent on their own parents for care. Having three generations under one roof did not benefit children, and under some circumstances actually harmed them. In a more recent paper, Campbell and Lee extended the analysis to other villages.(29) Among the phenomena observed in both villages was higher mortality for female orphans but not male orphans. This was attributed to reduced access to household resources on the part of the affected individuals. However, the Chinese data have serious shortcomings as many children who died were missing from the registers, especially if they were female, and precise dates for events were often not available.
The conclusion from this overview is that gender of the deceased parent and age of the child at the time of death indeed were important variables. Socioeconomic status of the family appeared to be a variable which had to be taken into account to ascertain whether indeed the status as orphan was responsible for increased mortality. Sex of the child only interacted with parental loss in the China-study.
The main shortcoming of the above-mentioned studies was that they did not take into account potential confounding factors. Information on several of these can however be collected and included, even in an analysis of 19th-century data. It is also essential to include duration since the loss of the parent in the study. In addition, the age ranges studied have to be rather narrow: in particular the crucial period before the first birthday of the child has to be adequately covered. By making use of a refined method of analysis and by including more detailed age-of-the-child and duration-since-parental-death information we try to add to historical knowledge about the importance of the father and mother role for the life of the child.
Orphans in the Netherlands: legal and social characteristics
Under Dutch law, anyone who was under the age of 23 was obliged to be under the authority of an adult - either a parent or a guardian. If one parent died, the surviving parent became the legal guardian of any minor children and a co-guardian was appointed by the magistrate.(30) A minor child who had lost both parents was appointed a guardian at the request of the minor's next of kin; if no request was made, it was the magistrate's responsibility to appoint a guardian. Before appointing a guardian, magistrates would seek the recommendations of four of the minor child's adult male relatives who could be either blood relations or adult males related by marriage.
Guardians were responsible for ensuring that minor children were cared for and educated but had no liability for the costs of their care or upbringing. Children who could not be placed with relatives and who had no-one else to care for them were put out to board with non-relatives, with the costs being paid by an institution for the relief of the poor.(31) The lower the cost to the Poor Councils or the Diaconate - the municipal and religious bodies committed to providing relief for the poor - the greater a child's chances of being boarded out.(32)
Orphanages played an important role in terms of providing care for orphaned children. In 1859, there were some 30,000 orphans in the Netherlands, 10,104 of whom were housed in 232 institutions.(33) As a rule, orphanages only admitted children who had lost both parents; children who had lost only one parent were either specifically excluded, or admitted only if their surviving parent was incapable of supporting or caring for them.
In most orphanages, the age limit for admission was between two and four years old. Children younger than this were boarded out until they had reached the required age. Burgher's orphanages and homes which were either wholly or partially privately funded also imposed their own admission criteria, on ideological grounds.(34) Because the aim of caring for children in burgher orphanages was to raise them to be citizens capable of earning their own living, sick, deformed or retarded children were often automatically excluded.(35) Admission to religious institutions was dependent on which church the child's parents had been a member of, and admission to burgher orphanages depended on which social class the child's parents had belonged to.(36) The municipality was responsible for caring for all children who could not be placed anywhere else.
The state of health in the large municipal orphanages which often accommodated hundreds of orphans, foundlings and abandoned or neglected children, was often strongly criticised.(37) Sub-standard accommodation facilities and atrocious hygiene conditions resulted in high mortality rates. Evidence of the poor state of orphans' health was also given by an inquiry into the physical condition of more than 16,000 young factory workers, schoolchildren and children in orphanages.(38) At the age of ten, the height, weight, lung capacity and muscle development of orphaned boys and girls were noticeably inferior to those of the schoolchildren and the children who worked in the wool factories. The survey concluded that "the physical development of orphans of a young age or upon arrival at the institution is severely retarded. The majority are amongst the neediest in society who have suffered deprivation and who are the offspring of parents who have in many cases passed on to their children the seeds of the diseases which caused their own premature demise. The fact that their subsequent development showed considerable improvement can be attributed to improved housing, clothing and nutrition."
By contrast, in private burgher orphanages, children were fed regularly and adequately, their clothes were well-maintained and frequently replaced, the accommodation was by no means sub-standard and their medical care also compared favourably with that of ordinary middle-class children. Prior to admission, all children were given a cowpox vaccination to prevent smallpox, and some orphanages employed a chirurgeon who provided the necessary medical care. The mortality rate in the burgher orphanages in the early nineteenth century was therefore much lower than the average mortality rate for the Netherlands as a whole.(39)
Woerden: a brief summary of the socioeconomic situation in the nineteenth and early twentieth centuries
Woerden was a typical provincial town, functioning as an important market for the surrounding countryside, having its own industry and a well-established middle class. At the beginning of the nineteenth century, Woerden had roughly 3,000 inhabitants; by the first few decades of the twentieth century, the figure had increased to more than 7,500. For many centuries, Woerden had been a brick and tile manufacturing centre, and the brickyards were a major force in Woerden's economy for a considerable time. However, as a result of internal and external competition and modernisation of' the production processes, the percentage of the labour force working in this sector declined from the middle of the nineteenth century on. A sharp rise in employment in the cheese trade, dairy farming and the manufacture of agricultural products occurred simultaneously.(40)
The region around Woerden is an archetypical orthodox-Protestant region. Around 1840, nearly 33 percent of Woerden's population was Catholic, almost 58 percent were members of the Dutch Reformed Church, 3 percent were Christian Reformed and nearly 5 percent were members of various smaller Protestant religious denominations.
The town and its region traditionally had a high mortality rate. Crude death rates (CDR) in the 1850s still reached an average value of 42 per 1000, of 34 per 1000 in the 1860s and of 40 per 1000 in the 1870s. Frequent epidemics, of cholera in particular, claimed numerous victims. In the mid-1860s, Woerden was hit by one of the last and most virulent epidemics of this sort.(41) When in the 1880s and 1890s an era of economic and social modernization started, the standard of living rose and epidemics were soon a thing of the past. The general level of mortality nonetheless remained rather high: it was only after 1892 that values of the CDR remained continuously below the 30 per 1000 level and only after 1904 that CDR remained below 20 per 1000. Infant mortality rates (IMR) in Woerden were extremely high. Published data for the second and third quarter of the nineteenth century show that IMR reached a value of 32.7 per 100 live births in 1841-1860 and of 34.3 per 100 live births in 1861-1874.(42)
Our main sources of information were the birth, death and marriage certificates for Woerden covering the period from 1850 to 1930 and the population registers covering the period 1850-1900.(43)
The collection of information on the life histories started with the marriage certificates. All marriages contracted during the period 1850-1899 were selected. For each married couple Christian names and surname, age, occupation, and municipality of birth and of residence were included in the database. If one of the spouses had been married before, this information was also included as was information on whether or not spouses were able to sign the marriage certificate. In the next stage, information on live births registered during the period 18501930 was linked to the information from the marriage certificates by making use of the surnames and Christian names of the parents, recorded on the birth certificate of the child. For each birth, information on the date of birth, sex, and Christian name of the child was included in the database. In the following stage information from death certificates for the period 1850-1930, including information on children registered as still-born, was linked to information on the married couple as well as to information on children born in this marriage. Again, linkage was done by using information on Christian names and surnames of the parents of the deceased and of the deceased themselves. Using information from marriage, birth and death certificates only implies that demographic events which occurred outside Woerden were not included in our database (exception made for deaths, as in case the deceased person was a resident of another municipality than the municipality of death, the registrar had to send a copy of the death certificate to the registrar of the municipality of residence of the deceased where it was included in the death register). For those marriages which had produced at least one birth in Woerden, information from the population registers of Woerden was added.
Continuous population registers were enforced in the Netherlands by the Royal Decree of December 22, 1849. The registers had to record the population residing within the municipality. The starting point for the first registers was the census of 1849. The returns from this census were copied into the population register, and from then on all changes occurring in the population in the next decade were recorded in the register. Each household was entered on a double page, with the head of the household first; he was followed by his wife (in case the head was a married male), children, relatives, and other members of the household. For each individual, date and place of birth, relation to the head of the household, sex, marital status, occupation, and religion were recorded. New household members arriving after the registration had started were added to the list of individuals already recorded, and those moving out by death or migration were deleted with reference to place and date of migration or date of death. In fact the population register combined census listings with vital registration in an already linked format for the entire population. Families and individuals can, in principle, thus be followed on a day-by-day basis for a long period. In Woerden the first register covers the period 1850-1854. In addition to this register, information from the registers for the periods 1855-1880 and 1881-1900 was used. The main pieces of information from the register were religion, dates of birth and dates of migration of families.
In our analysis, attention is primarily directed to the survival prospects of the child in four different family contexts; the mother of the child has died, the father has died, the surviving parent has remarried and the child lives in a complete family. Given the fact that the number of families in which both father and mother had died while the child was aged less than twelve was extremely small, we did not analyse this family situation separately.
We already referred to the fact that the effect of parental loss on infant death could be a statistical artifact. For example, parent and child could live under common conditions that had negative health consequences for the parent and also contributed to lowered survivorship of children. Clear examples are socioeconomic status (because social class is negatively associated with higher mortality, the higher death rates of orphans could reflect social class differences) and epidemics. We take these statistical artifacts into account by including these variables into a multivariate analysis.
We first of all expect that the effect of parental loss on child mortality is dependent on the socioeconomic status of the family. Variations in social status determine the exposure and resistance to health-threatening factors: the availability of food, quality of water, clothing and bedding, size and quality of housing, means to purchase materials needed for hygienic and preventive care etc. As a measure of the socio-economic position of the family we use the occupational information relating to the (deceased or surviving) father. The occupation of the father not only measures the position of the widower, it also gives an indication of the position of the surviving widow. In general one might state that widows from higher and middle classes were either well-off in their own right, inherited property from their husbands, or were able to take over ownership and operation of their husbands' trade and business. Sometimes, they were also supported by pensions from life insurance companies or state-sponsored widows' funds. Most widows from social classes that did not own property were dependent on benefits from organized charity or had to find work in low-paid, home-based employment, like washing, knitting, and sewing, that made it possible to combine an occupation with child and home care. Our occupational classification is partly in line with the division described here. The occupation of the father was classified into six groups: Upper class (employers in industry, professionals, high civil servants; higher military); Petty bourgeoisie (shopkeepers, small entrepreneurs and merchants; self-employed artisans); Lower level professionals and lower civil servants; Farmers, fishermen, hunters, agricultural labourers; Skilled manual workers (craftsmen in small business; craftsmen and skilled labourers in industry; service employees and lower military); Casual and unskilled labourers (casual labourers; construction workers; unskilled labourers in crafts and industries); and finally the group Without occupation.
Both parents and children may have run increased risks of dying during periods characterized by excess epidemic mortality because they were temporarily subjected to the same risk factor. It is clear that the strong association which in this situation can be observed between parental loss and infant and child mortality is not an effect of the child being orphaned but caused by the fact that parents and children shared a common hazardous environment. To take this factor into account, we included in our analysis a variable which indicated whether the child experienced such a high-mortality period. Three periods of excessive levels of mortality were distinguished: 1858-1859 (smallpox and cholera), 1866 (cholera) and 1870-1871 (smallpox).
In addition to the other mentioned factors, period (birth year of the child) is included in the model. This is among other things related to the fact that during the period 1850-1900 important changes in mortality took place as a consequence of which children who lost their parents during the end of the nineteenth century ran lower death risks than children who spent the first twelve years of life in the middle of the nineteenth century. As the levels of infant and childhood mortality during the period studied were characterized by first an increase and later on by a decrease we also included the squared value of the birth year of the child in the analysis to take into account the parabolic relationship between mortality levels and year of birth.(44)
Many studies have shown that age of the mother is strongly correlated with infant, and in particular with perinatal mortality. Age of the wife also has an effect on her own mortality risks and as age of men and wife also are strongly interrelated, age of the wife is also correlated with the husband's mortality risk. For that reason, age of the wife at birth was also included in the model. The (perinatal) mortality risks associated with age of the mother are generally U-shaped, mortality among young and old mothers being at a maximum.(45) To take into account the parabolic relationship between age of the mother and mortality of the child, age as well as the squared value of age was entered into the equation.
The risks of neonatal death are generally greater with the first birth than with the second, and thereafter increase with third, fourth, and fifth or more births. High birth ranks, and large numbers of children in the family usually result in short birth intervals. As more babies are born at smaller intervals, the children already present have a greater chance of being neglected and the penultimate child will be deprived of breast-feeding because of a new pregnancy with all its consequences. In their turn large numbers of children also lead to higher death risks for the mother. Regardless of maternal age, increasing parity usually led to an increased risk of maternal mortality too.(46) The birth rank of children from the sample could be determined by making use of the information on date of birth of the child from the birth certificates and the information from the population registers. As only those children were included which were born in Woerden or who were included in the register, the parity of the child as deduced from these sources may be lower than the real parity of the child.
Several other factors, which do not necessarily affect both parent and child at the same time, but are known as risk factors for infant mortality were also included in the analysis. This relates to religion, literacy and a crude indicator of the availability of social support.
Religious traditions, norms, and attitudes - beliefs about disease causation, culturally conditioned patterns of dietary intake of mothers and children during pregnancy, lactation, weaning, and illness - shape the health-related practices of individuals and influence the health and survival of adults and their children.(47) A Dutch study dating from the beginning of the twentieth century showed that Dutch Catholics had a particularly high infant mortality. One aspect which was often associated with the high Catholic infant mortality rate was their low proportion of breast-feeding. More than other groups, Catholic women were expected to accept the burden and the risks of motherhood. At an extreme, in case the life of the mother was in danger as a consequence of her pregnancy or delivery, abortion for medical reasons to save the life of the mother was not allowed.(48) To take the effect of religion on infant mortality into account, we differentiated the families in our sample in three groups: Roman Catholic, Protestant and Other (including Unknown and Without religion).
Numerous studies have shown that parental education, and particularly maternal education, has a large impact on survival and health of children. This effect can only partly be accounted for by the economic advantages associated with higher levels of education such as higher income and higher housing quality, factors which we tried to capture by the inclusion of the social status of the family head. Cleland and Van Ginneken concluded that, compared to the uneducated, educated mothers attach a higher value to the health of children, are less fatalistic about disease and death, are more knowledgeable about disease prevention and cure, are more innovative in the use of remedies, and are more likely to adopt new codes of behaviour which improve the health of children.(49)] To take the potential effect of the educational level of the mother into account, we used information on her ability to sign the marriage certificate. Around 25 percent of the wives were not able to sign.
The effect of the loss of the parent is assumed to be partly dependent on the availability of a support network which may buffer against stress. Members of the social network might provide assistance to lone parents with child-care tasks and responsibilities. Social networks also afford the context for collective standards and therefore for the enforcement of child-care norms. Embeddedness of child-rearing in kin and community acts against the social isolation that has been linked with child maltreatment.(50) When surviving parents can rely on kin for childcare, housekeeping, care during illness, emotional support, or the provision of food and shelter, parents and children are better off and the child's vulnerability is decreased.(51) The accessibility of the family network and the opportunity to call upon them when immediate care was necessary was assumed to be greater for native-born people.(52) To indicate whether or not a surviving parent and a child could fall back on the support of members of his or her family, we used a crude indication of the degree to which the parents could be considered as 'settled' inhabitants of Woerden. We differentiated between families in which both spouses were born in Woerden or the adjacent villages, families in which the husband was born in these municipalities, families in which the wife was born in the selected municipalities and other families.
Finally, as a recent study in Bangladesh showed that the loss of the father was associated with an increase in the child mortality rate, regardless of the child's gender, whereas the mother's death was associated with almost a 200 percent increase in the mortality of her sons and a more than 300 percent increase in the mortality of her daughters, we also included sex of the child as a variable, to study possible greater vulnerability of daughters than sons.(53)
To summarize, we were able to measure the effects of mortality of the parent(s) on the life chances of the child in relation to the sex of the child, its age at the death of one or both parents, the duration since the death of the parent, the occupation of the father, the level of schooling, the age of the mother, the period of birth of the child, the birth rank of the child, place of birth of parents and religion of the parents. Information on the presence of persons other than siblings in the household, or on the financial position of the widow(et) was not included.
For each child, age at death was calculated in days; where this age was not available, information on age at date of departure was available. As we were primarily interested in effects of the loss of the parents during the period that the child was still dependent on its parents, we only studied mortality of children during the first twelve years of life.
The data file constructed for this analysis is a child-level file, i.e. each record is a child with the mother's and father's information attached. Since the observations are live births, there is frequently more than one observation per family and these observations are not independent. The statistical effect of ignoring the correlation within families would be incorrect estimates of the standard errors of the parameters (underestimation), leading potentially to incorrect references. Given the fact that the ages of the children at the time of death of father or mother differed considerably and that the effect of parental loss depended on the age of the child, and more generally, that there is wide variance in personality traits among children raised in the same family, as a consequence of which each of them is experiencing the death in his or her own manner, we nonetheless feel confident in our treatment of the children as individuals.(54) A total of 937 families were studied, who had in total 3936 live-born children.
Event history modelling is used to examine the impact of family history on the risk of dying of the child. In such modelling techniques, the rate or hazard rate at which an event, in this case death, occurs is modelled.(55) A hazard rate refers to the probability that an event occurs within an infinitesimally small interval, given that this event has not occurred before the start of that interval. The relationship between the hazard rate, time, and a set of covariates (or independent variables) is modelled by the following expression:
[r.sub.(t,X)] = [[h.sub.(t)].sup.*] [exp.sup.bX],
where [r.sub.(t)] is the death rate at age t for children with covariate vector X, [h.sub.(t)] is the so-called baseline rate at age t, and X is a vector of covariates.
Hazard analysis allows us to overcome some of the limitations of other, less suitable, approaches to modelling the impact of independent variables on the probability of death among children. Logistic regression is not suited, because it only models whether or not a child has died during childhood, without accounting for the impact of age on the probability of dying. Ordinary least square regression is not suited, because one either has to drop those children from the analysis who have not died during childhood, or has to assign an arbitrary age of dying to them. Hazard analysis is superior to both techniques, because it makes optimal use of information on (a) whether or not a child died, and (b) the age at which a child died. Statistically, this is done by maximizing a likelihood function consisting of two parts, viz. a density function f(t) for those children who died at age t, and a survival function G(t) for those children who survived until age t.
Another reason why hazard analysis is superior to other techniques in analysing events, is its ability to model the impact of independent variables that change their value during childhood. For instance, a child's mother might die when the child is three years old. If so, the independent variable 'mother alive' changes its value from 1 to 0 at age three. Such instances are handled in hazard analyses by splitting the period during which a child is observed in two parts: A first part that runs from birth until the date at which the mother died, and a second part that runs from the date at which the mother died until the date at which the child itself died, or was lost from observation because it became twelve years of age or migrated.(56) An independent variable that can change its value during the period of observation is called a time-varying covariate.
A third advantage of hazard analysis is its flexibility in analysing different types of effects of independent variables. For instance, one could assume that the impact of a mother's death on the death rate of her child simply depends on whether or not she is alive. Such an effect is called a state-dependent effect. However, one could also assume that the impact of the death of a mother is stronger shortly after her death than later on. Such an effect can be called a duration-dependent effect. One way to model this is by specifying two time-varying covariates, one indicating if the mother died less than six months ago, and one indicating if the mother died more than six months ago. A third possibility is to assume that the impact of a mother's death depends on the age of the child. Such an effect is called a time-varying effect. This type of effect can be modelled either by specifying an interaction between the death of the mother and the age of the child or by estimating separate models for young ages (e.g. the first six months of a child's life during which the risk of death is by far the highest) and older ages (e.g. between six months and twelve years) and comparing the effects of the variable 'mother alive' between the two models. As these examples illustrate, hazard analysis allows for truly dynamic modelling of the impact of family structure on the risk of dying during childhood.
Hazard models differ among themselves in the way in which the age or time dependence of the process under study is modelled. In so-called parametric models, this age dependence is modelled by specifying a parametric form for the baseline rate ([h.sub.(t)]). For instance, if the death rate is assumed to decline monotonically during childhood, the Weibull or Gompertz function can be chosen to model the age dependence of the process. A disadvantage of parametric models is that misspecification of the functional form of the age dependence can bias the estimates of the effects of the covariates in the model. Because our main interest is in the effect of these covariates (in particular those covariates related to changes in family structure) on the death rate, rather than on age-related variations in the death rate, a so-called semi-parametric hazard model is estimated. In this model, introduced by Cox, only the effects of the set of covariates is estimated, but the shape of the baseline rate ([h.sub.(t)]) is left unspecified.(57) A possible disadvantage of this Cox-model is that it assumes that the effects of the covariates are time-in-varying. To counter this disadvantage, a two-stage procedure is followed. In the first stage, a Cox-model for the whole period between childbirth and age 12 is estimated. In the second stage, separate Cox-models for the first six months since birth, and for the period between six months and twelve years are estimated. By comparing the estimates from these models, one can conclude whether or not the impact of covariates depends on the age of the child.
In all models, the strength of effects is expressed in terms of relative risks. The interpretation of such relative risks depends on whether the independent variable is categorical or continuous. For categorical variables, it expresses the increase or decrease in the death rate compared to some reference category. For instance, a relative risk of 1.35 for boys implies that boys have a (1.35-1)*100 percent = 35 percent higher death rate than girls. For continuous variables, it expresses the increase or decrease in the death rate given by a one-unit change in the independent variable. For instance, a relative risk of 1.03 for 'parity of child' implies that the death rate of children increases with (1.03-1)*100 percent = 3 percent for each higher parity. The statistical significance of individual coefficients can be determined by t-tests, where t-values of 1.96 or more will be considered to point to statistically significant differences.
All model estimations have been done with the computer program TDA (Transition Data Analysis), developed by Gotz Rohwer.(58)
As could be expected, children in our database were characterized by very high mortality. During the second half of the nineteenth century almost 29 percent of the live born children did not reach the end of the first year of life. Age five was not reached by almost 38 percent of the children born during the period 1850-1899 whereas at age twelve, the age to which we restricted our analysis, more than 40 percent of the children had died. In particular during the first half year mortality was very high. It must be rembered that by excluding all children which had died before a birth certificate was made out, some of whom thus were incorrectly registered as stillborn, we in fact even underestimate mortality levels.
In the first stage of the analysis, we tested the hypothesis that the effect of parental loss on child mortality varied only according to gender of the surviving parent. Table 1 presents parameter estimates for the effect of parental loss on the mortality rates of children. Four situations were distinguished: the original family was still complete, the father had died and the widow had not remarried, the mother had died and she was not replaced, and finally, a situation in which one of the parents had died and the surviving parent had remarried. The relative risks indicate in this case the difference in mortality rates between children living in any of the last three situations in comparison with children in a complete family. As already mentioned, families in which both father and mother had died while the child was aged less than twelve were not analysed separately.
Table 1 shows clearly that children living in a broken home had much higher mortality rates than children both of whose parents were still alive. In particular, the loss of the mother brought with it an increase in mortality: children's death risks were more than two times higher than when father and mother were still alive. The effect of the loss of the father was much lower although here too it had negative consequences for the survival prospects of the child. Remarriage of widow or widower did not result in a death rate which differed significantly from those of children in complete families. This may be related to the association between the time between parental loss and remarriage. We expect the strongest effect of parental loss directly after the death of the parent; usually several months elapsed between death and remarriage of the widower, and during this most critical period only few remarriages took place. For women, the effect of remarriage may even have been more difficult to observe. Remarriage among women was even much less frequent at short durations of widowhood - when the highest excess mortality is expected - due to the fact that a mourning period of 300 days was stipulated for them.
In line with the expectations, we observed that boys had significantly higher death risks than girls. The year in which the child was born (and during which it spend the most risky period of its life) had a strong effect on mortality rates: as 1850 was used here as the reference year, the two time variables indicate that [TABULAR DATA FOR TABLE 1 OMITTED] after an initial period of rising mortality, death rates started to decline again. The inclusion of a variable indicating whether or not a child had lived during a year in which Woerden was struck by excessive epidemic mortality appeared to be worthwile as this variable had a very strong effect on death risks of children. The birth rank of the child also had an effect in the expected direction with higher parities showing higher death risks. The educational level of the mother, measured through her ability to sign the marriage certificate, had a strong effect on child mortality: children born out of these mothers had an 18 percent lower mortality risk. Independent of this, the socio-economic status of the family also affected mortality rates: compared with the reference group of unskilled labourers, all social classes had lower infant mortality. Children born in families of shopkeepers or small artisans (petty bourgeoisie) and children from skilled labourers had mortality risks which were more or less comparable with those of the reference group whereas farmers, the upper class and lower civil servants were characterized by lower mortality. Contrary to our expectations, children from Roman Catholic families had lower mortality than children from a Protestant background. This might be related to the fact that one factor through which religious affiliation affected the infant mortality rate - the low proportion of breast-feeding among Catholics - was no longer able to differentiate the various groups as soon as the mother had died. It might also be caused by the fact that the orthodox Protestant community of Woerden deviated from mainstream Dutch Reformed Protestantism, which usually has been the group with which Dutch Catholics compared so unfavourably. Rutten has recently shown that strict Calvinists in the province of Zuid-Holland indeed were characterized by completely different attitudes towards for example 19th-century vaccination campaigns.(59)
In the second stage we investigated whether the effects of parental loss differed according to the time elapsed since the death of the parent. In a preliminary analysis, we tested a refined model in which we differentiated mortality risks of children according to four durations since the death of the mother: families in which the mother had died less than one month before, families in which the mother had died between one and 6 months before, families in which between 6 and 12 months had passed since the death of the mother and finally families for which more than one year had passed since the loss of the mother. It appeared that relative risks in the first two situations hardly differed (respectively 3.97 and 4.46) and that t-values in both cases were highly significant. Children whose mothers had died during the previous 6-12 months or who had died more than one year before, also had more or less the same relative risks (1.65 respectively 1.10), both of which did not deviate significantly from 1.00. We therefore decided to group durations below 6 months together and did the same with durations above 6 months. The results of the analysis of mortality risks of children by duration since death of the father or mother are given in table 2.
The table clearly shows that the time elapsed since the death of the parent had a very strong effect on the mortality risks of the child. Whereas children whose mother had died only recently had a mortality risk which was more than four times higher than that of children living in complete families, children whose mother had died more than 6 months before had no significantly increased mortality risk. For fatherless children the situation was comparable, although the effects of the loss of parent were indeed much smaller: children who had lost their father less than six months before had a more than two times higher mortality risk than children in complete families. For children living in a fatherless family in which the father had died more than 6 months before, no significantly increased mortality risk was visible. For children living with a stepfather or stepmother mortality risks were not significantly different from those of children living in a complete family. The effect of the other mentioned variables was already discussed earlier and did not deviate from the situation described in table 1.
As was stated before, we expected that the effect of parental loss on child mortality was not only dependent on the gender of the parent and the duration elapsed since the death of the parent but also on the age that the child had reached when experiencing this loss. Given the shape of the survival curve, the restricted number of cases and the fact that breast-feeding, and thus the presence [TABULAR DATA FOR TABLE 2 OMITTED] of the wife, is of much more importance during the first half year in the life of the child than during the period thereafter, we distinguished two periods: children aged less than 6 months at the time of death of the parent and children aged 6 months or more. Of course it is impossible to analyse the joint effects of age of the child and duration since loss of the parent for children aged less than 6 months when we keep the earlier distinction between durations since parental loss of less than 6 and more than 6 months. Therefore, we restricted our analysis of age of the child at the time of death of the parent in combination with duration elapsed since the loss of the parent to children aged 6 months or more when losing their parents. Furthermore, children living in step families were treated as children living in complete families as the two family situations showed no significant differences in survival prospects of children. Results for the youngest age group are given in Table 3, for the age group 6 months and older in Table 4.
Table 3 shows that, during the first six months of their lives, the death rate of children whose mother had died was almost 6.5 times higher than for children in complete families. A significantly higher risk was also observed for children of which the father had died. Relative risks here were almost 2.5 times higher than for the children acting as reference group. It appeared that for children aged less than 6 months the socio-economic position of the family did not have a statistically significant effect on survival, whereas the educational level of the mother still had a decreasing effect on mortality risks. Increased survival was also associated with the child being a girl, low birth rank, and being Roman Catholic. Epidemic years were no longer associated with increases in mortality when it came only to very young children.
[TABULAR DATA FOR TABLE 3 OMITTED]
[TABULAR DATA FOR TABLE 4 OMITTED]
In comparison with the youngest age group, children aged 6 months or over at the time of death of the parent showed a completely different pattern. Whether or not the mother was still alive no longer had a statistically significant effect on mortality of the child! Even for children which had lost their mother very recently (i.e. less than 6 months ago) no statistically significant effect on mortality was observed. Parental loss on the other hand still had a statistically significant effect on survival of children when it was the father who had died and when death had taken place only very recently (less than 6 months prior). When the death of the father had occurred more than 6 months before, children aged 6 months or over did not undergo an increased risk of dying compared with children living in complete families. The mortality risks of children aged 6 months or over did not differ anymore according to gender. However, the socio-economic situation of the family and the educational level had a statistically significant effect on mortality: children from higher social classes and with an educated mother showed lower mortality risks. Epidemics were important as far as mortality in the age group 6 months-12 years was concerned. Furthermore, the increase and later on the decrease of mortality by period as well as the unforeseen direction of the influence of religion were observed here too.(60)
Conclusions and suggestions
Our study has shown that during the nineteenth century the loss of one of the parents seriously endangered the physical well-being and the survival prospects of children. This effect was not a statistical artifact: neither specific risky events shared by parents and children (such as an epidemic) nor a common unfavourable environment shared by parent and child (such as low social class) could explain this relationship.
In general children were worse off when they had lost their biological mother than when they had lost their biological father. The effects of gender of the surviving parent varied however considerably according to the age of the child at the time of death of the parent, and according to the time which had passed since the death of the parent.
In particular during the first months after delivery maternal death could lead to infant death very soon due to the lack of adequate feeding of the child. The results of earlier studies, which showed that depriving the new-born child of the person who could provide it with adequate feeding, thus were clearly confirmed. During the first half year of the life of the child, other activities of the mother such as food preparation, washing clothes, bathing the child, house cleaning and sickness care of course also could affect the survival of the child. But whereas there was hardly any alternative for breastfeeding and enormous hazards were associated with artificial feeding for all social groups, activities like child care, sanitation measures and control of hygiene could be taken over by other (female) family members or neighbours or were partly dependent on the socio-economic position of the family. The fact that even after controlling for the socio-economic position of the family and the availability of kin, the effect of maternal loss remained so strong thus clearly indicates that in particular the practical impossibility of finding substitute food for the new-born was the most important factor.
Although the effect was less far-reaching, the loss of the father also had negative consequences when the child was still very young. This might be interpreted as an indirect effect of the death of the bread-winner. Part of the mother's time now was required for economically productive activities, and less time (or a shorter duration) was available for breast-feeding, and for maternal activities. In particular premature weaning of the child due to the necessity to enter the labour force may have played a role.
Remarkably enough the role of the father was more important than that of the mother when the child had survived the first six months of his life, in particular during the first 6 months after the death of the father. This showed itself in the increased death risk of fatherless children, aged 6 months or over at the time of death of their father. Families then were dependent on the household head as provider of goods and children could no longer depend only on breastfeeding. The death of the father could at least temporary lead to a situation in which the level of consumption of children decreased so much that health was affected adversely. The household's attempt to cope with the death of the male adult may have shifted labour away from health maintaining activities such as cleaning, collecting water, hygienic food preparation, and breast-feeding. Again, premature weaning may have been an important factor. The fact that this increasing effect on child mortality lasted only several months indicates that mother and children were able to adapt themselves to the new situation very quickly. This conclusion can be generalized to other situations as well: while the death of a parent may have put a child at risk, this risk was not sustained over time. The majority of children and their surviving parent adapted successfully to the new circumstances and learned their new roles after the first year.
Viewed from the standpoint of the physical well-being of the child, remarriage of the surviving spouse was an efficient solution for the problems with which they were confronted. Our results confirm those of Swedish researchers, and indicate that children whose parent remarried had mortality risks which did not deviate from those of children living in complete families. Stepchildren were thus not so vulnerable that they were running higher mortality risks than children in complete families and at least as far as survival prospects is concerned, children getting a step-parent were certainly better off than those children living in a one-parent family. Growing up in a step-family thus appeared not to be as harmful as often is suggested by folk belief and fairy-tales, and the selective neglect of stepchildren, which has been mentioned in contemporary studies, did not in our historical population result in greater mortality risk.
After the child had reached a certain age, orphanhood appeared no longer to have significant effects on survival prospects. Earlier Swedish research showed contradictory results. What is worth noting in this context is that children in orphanages, who usually were admitted at ages between two and four years, were depicted as having a retarded physical development upon arrival at the institution. The questions which arise regarding the value of mortality as a measure of physical health, regarding selection processes for admission to orphanages and the social background of the admitted children, and regarding the effect of living in an institution are numerous but for the moment cannot be answered.
We have to stress that the results of our study are based on a rather restricted data set. That relates not only to the number of observations, but also to the variables which we were able to include in the analysis and to the way the data were analysed. First of all, we were not yet able to use all information from the population registers. In particular information on the composition of the household, the changes therein and the relation between the members of the household was not included completely. Therefore, specifying the family situation of the child in its full complexity was not possible. Hidden caretakers, like aunts, uncles, grandparents could not yet be identified.
Secondly, more precise information on the socio-economic situation of the household before and after the death of the parent has to be collected to study whether changes in income were responsible for increased death rates of children. Information on income taken from the registers of the hoofdelijke omslag, or local income tax (1849-1910), could be used to derive changes in the income situation of the surviving parent.
Directly related to the socioeconomic and health situation of the child is also the mother's participation in the labour force.(61) In other studies, working status of the mother was a variable which distinguished widows who succeeded in their children's survival from those women who did not. When the mother's labour was also required for economically productive activities, because there was no husband present who could provide support, women could spend less time on maternal activities. Unfortunately, the quality of the historical information on women's participation in the labour force given in the population register is usually very poor. We do not know whether the mother worked exactly at the time her husband died, or only before or after that point. Nevertheless, an effort must be made to make more of this information than has been done here.
Given the small number of observations, extending our analysis over a longer period of time, and including information on children born in Woerden from marriages contracted outside the town, has a high priority. It would give more confidence in the results of our study, and could make more detailed analysis possible for example of the effect of remarriage, the interaction between period of birth of the child and survival prospects etc. At the same time, it would make it easier to apply more refined techniques of analysis. Carrying out an analysis which takes into account the interrelatedness of the observations on children from the same family for example by using multilevel models might lead to more correct estimates of the standard errors of the parameters of the Cox model.(62) It might make it possible to model and to investigate at the same time the influence of characteristics of the family (income, socio-economic status) as well as those of the individual children concerned (age, sex etc.).
To gain more definitive answers on the relative importance for survival of the roles that fathers and mothers played during the first years of the life of the child, more in-depth data are also needed. One might for example think of a more complete description of the family network and the role kin members played in supporting the surviving parent and its child, and of a more detailed account of the economic situation of the widow and widower. For that purpose, the data from our study could for example be linked to the information available on the decisions taken by magistrates regarding the appointment of guardians and co-guardians. In addition to that, by comparing the life course of orphans who were boarded out with non-relatives with that of (partial) orphans who were placed with relatives or were admitted to institutions, the effect which institutionalization had on the life of the child could be studied in a more satisfactory way.
P.O. Box 11650 2502 AR The Hague Netherlands
This is a revised version of a paper presented at the 22nd Social Science History Association Conference (Washington DC, 1997) and at the Ecole des Hautes Etudes en Sciences Sociales (Paris, 1998).
1. Wendy Post, Frans van Poppel, Evert van Imhoff and Ellen Kruse, "Reconstructing the extended kin-network in the Netherlands with genealogical data: problems, methods and results," Population Studies 51 (1997): 263-278.
2. Ida Blom, "The history of widowhood: A bibliographic overview," Journal of Family History 16 (1991): 191-210.
3. Stephen Collins, "British stepfamily relationships, 1500-1800," Journal of Family History 16 (1991): 331-344.
4. Lourens Roelof Kooij, Eenoudergezinnen in de huisartspraktijk (Lelystad, 1988).
5. Pekka Martikainen and Tapani Valkonen, "Mortality after death of spouse in relation to duration of bereavement in Finland," Journal of Epidemiology and Community Health 50 (1996): 264-268.
6. Margareth S. Stroebe and Wolfgang Stroebe, "The mortality of bereavement: A review," in: Margareth S. Stroebe, Wolfgang Stroebe and Robert O. Hansson eds, Handbook of bereavement, Theory, research, and intervention (Cambridge, 1993) 175-195. For examples relating to tuberculosis, leprosy, malaria, syphilis, see Mead Over, Randall P. Ellis, Joyce H. Huber and Orville Solon, "The consequences of adult ill-health," in Richard G. A. Feachem et al. Eds., The health of adults in the developing world (Oxford, 1992) 161-208.
7. Stroebe and Stroebe, "The mortality" 178-183.
8. Stroebe and Stroebe, "The mortality" 188-192.
9. Phyllis R. Silverman and J. William Worden, "Children's reactions to the death of a parent," 300-316 in Stroebe, Stroebe and Hansson eds., Handbook of bereavement 300-316.
10. G.W. Brown, T. O. Harris and A. Bifulco, "Long-Term effects of early loss of parents," in Michael Rutter, Carroll E. Izard and Peter B. Read eds., Depression in young people. Developmental and clinical perspectives (New York/London, 1986) 251-296.
11. W.H. Mosley and L.C. Chen, "An analytical framework for the study of child survival in developing countries," in W.H. Mosley and L.C. Chen eds., Child survival: strategies for research, Population and Development Review, 10 (1984): 25-45.
12. Silverman and Worden, "Children's reactions" 311-312.
13. About these stereotypes see Frans van Poppel, Trouwen in Nederland. Een historischdemografische studie van de 19e en vroeg-20ste eeuw (Wageningen, 1992) 308-311. About problems related to changes in role functioning see Stephen R. Shuchter and Sidney Zisook, "The course of normal grief," in Stroebe, Stroebe and Hansson eds., Handbook of bereavement 23-43.
14. Shuchter and Zisook, "The course" 38-42; Stroebe and Stroebe, "The mortality" 185.
15. Marjo Buitelaar, "Widow's worlds. Representations and realities," in Jan Bremmer and Lourens van den Bosch eds., Between poverty and the pyre. Moments in the history of widowhood (London, 1995) 1-19.
16. Silverman and Worden, "Children's reactions" 302; Brown, Harris and Bifulco, "Long-term effects".
17. Jill Kornin, "Child maltreatment in cross-cultural perspective: vulnerable children and circumstances," in Richard J. Gelles and Jane B. Lancaster eds., Child abuse and neglect. Biosocial dimensions (New York, 1987) 31-55.
18. A. Rochester, Infant mortality: results of a field study in Baltimore, Maryland. based on births in one year (Washington, 1923) 115. See also R.M. Woodbury, Infant mortality and its causes (Baltimore, 1926) 43 and J. Yerushalmy, M. Kramer, and E.M. Gardiner, "Studies in childbirth mortality: Puerperal fatality and loss of offspring," Public Health Reports 55 (1940).
19. Silverman and Worden, "Children's reactions" 306-312.
20. Silverman and Worden, "Children's reactions" 302.
21. Margareth S. Stroebe, Robert O. Hansson and Wolfgang Stroebe, "Contemporary themes and controversies in bereavement research," in Stroebe, Stroebe and Hansson eds., Handbook of bereavement 457-476.
22. Peter Uhlenberg, "Death and the family," Journal of family history 5 (1980): 313-320.
23. Samuel H. Preston and Michael R. Haines, Fatal years. Child mortality in late nineteenth-century America (Princeton, 1991) 123-127.
24. Magdalena Bengtsson, Det hotade barnet. Tre generationers spadbarns- och barnadodlighet i 1800-talets Linkoping (Linkoping, 1996).
25. Ulf Hogberg and Goram Brostrom, "The demography of maternal mortality: seven Swedish parishes in the nineteenth century," International Journal of Gynecology and Obstetrics 23 (1985): 489-497.
26. Sune Akerman, Ulf Hogberg and Tobias Andersson, "Survival of orphans in Nineteenth-Century Sweden," in Lars-Goran Tedebrand ed., Orphans and foster-children. A historical and cross-cultural perspective (Umea, 1996) 83-103.
27. Birgit Persson and Lisa Oberg, "Foster-children and the Swedish state 1785-1915," in Tedebrand ed., Orphans and foster-children 51-81.
28. Cameron Campbell and James Z. Lee, "A death in the family: Household structure and mortality in rural Liaoning: Life-event and time-series analysis, 1792-1867," The history of the family 1 (1996): 297-328.
29. Cameron D. Campbell and James Z. Lee, Price fluctuations, Family Structure and Mortality in Two Rural Chinese Populations: A Comparison of Peasants and Serfs in Eighteenth and Nineteenth-Century Liaoning: Paper presented at the IAEH/IUSSP workshop, Population and Economy: From Hunger to Modern Economic Growth (Osaka, 1997).
30. J. van Koetsveld van Ankeren, Handleiding voor voogden en toeziend voogden, volgens het Nederlandsch Wetboek (Assen, 1839).
31. D.Q.R. Mulock Houwer, "De wezenzorg in de 19e en 20e eeuw," in B. Kruithof, T. Mous and Ph. Veerman eds. Internaat of pleeggezin. 200 Jaar discussie (Utrecht, 1981) 43-64.
32. An important characteristic of the Dutch system of care for the poor was the incorporation of private mainly denominational charity in the legal structure. The 1854 Poor law only authorized the civil authorities to act when private authorities were not able to fulfil their duties towards the poor. In each municipality the Protestant churches had a deaconate which was responsible for providing relief for the poor, for the boarding out of children in families, for the admission of orphans to the orphanages and for their livelihood. Each Roman Catholic parish also had its own poor council, responsible for the care for orphans whereas at the local level a Catholic orphanage was functioning. See H. Smissaert, Overzicht van het Nederlandsch armwezen (Haarlem, 1910).
33. The outcomes of the censuses as far as the population living in institutions is concerned were considered not to be very reliable and at least not very well comparable over time and region. See CBS, Inleiding tot de uitkomsten der Negende Algemeene Tienjaarlijksche Volkstelling met daaraan verbonden Wonlng- en Beroepstelling ('s-Gravenhage, 1913) 65.
34. 'Burghers' were people who lived in towns and cities who had civil rights by virtue of their social status, wealth and/or duration of residence. The various charitable institutions that they had funded, including orphanages for children from similar, privileged backgrounds were the 'burgher' orphanages.
35. J. Dankers and J. Verheul, Als een groot particulier huisgezin. Opvoeden in bet Utrechtse Burgerweeshuis iussen caritas en staatszorg 1813-1991 (Zutphen, 1991) 78; J.J. Dankers, Wezenzorg en liefdadigheid. Van Utrechts Gereformeerd Burgerweeshuis tot stichting her Evert Zoudenbalch Huis, 1813-1991 (Utrecht, 1991).
36. S. Groenveld, J.J.H. Dekker, and Th.R.M. Willemse, Wezen en boefjes. Zes eeuwen zorg in wees- en kinderhuizen (Hilversum, 1997) 269-274.
37. Groenveld et al., Wezen en boefjes 300-301.
38. Rapport der Commissie belast met het onderzoek naar den toestand der kinderen in fabrieken arbeidende ('s-Gravenhage, 1869).
39. J.L. de Jager, In een ander thurs. De pedagogische geschiedenis van bet R.C. Jongensweeshuis en Amstelstad in Amsterdam (Amsterdam, 1985) 151. See also Anne McCants, "Nederlands republikanisme en de politiek van liefdadigheid", Tijdschrift voor sociale geschiedenis 22 (1994): 443-455, for a comparison of life expectancy in poor and richer Amsterdam orphanages.
40. Rob van der Laarse, Bevoogding en bevinding. Heren en kerkvolk in een Hollandse provinciestad, Woerden 1780-1930 ('s-Gravenhage, 1989) 32-34; 236-240.
41. For the period studied here, in particular the cholera epidemics of 1853-55, 1859, 1866 and 1882, are worth mentioning. In addition to that, in 1857 and 1858 and in 1871 and 1882 outbreaks of smallpox took place, whereas in 1879 whooping cough and in 1876 and 1884 measles and in 1901 and 1918 diptheria prevailed. See Nico Plomp, Ziekenzorg in Woerden (Woerden, 1980) 77-96.
42. Sterfte-atlas van Nederland, 1841-1860, Nederlandsche Maatschappij tot bevordering der Geneeskunst (Amsterdam, 1866); Sterfte-atlas van Nederland over 1860-1874 (Amsterdam, 1879).
43. Although the regents of the former Woerden orphanage - the Municipal Orphanage or Reformed Orphanage - were still responsible for the orphans, Dutch Reformed orphans were sent to live with families of labourers and artisans from 1811 onwards. The Catholic orphanage founded in 1804 had by 1882 also ceased to be used for its original purpose and some of its orphans were sent to live in a Home for Gentlewomen where they were looked after by the horne's live-in baker. See G.N.M. Vis, Het weeshuis van Woerden. 400 jaar Stadsweeshuis en Gereformeerd Wees- en Oudeliedenhuis te Woerden (Hilversum, 1996) 174-176. In view of the small numbers of children admitted to orphanages and the lack of any clear institutional structure, our research will not devote any special attention to orphaned children from Woerden placed in orphanages.
44. Frans van Poppel and Kees Mandemakers, "Differential infant and child mortality in the Netherlands 1812-1912: First results of the historical sample of the Population of the Netherlands," in A. Bideau, B. Desjardins, and H. Perez-Brignoli, Child and infant mortality in the past (Oxford, 1997) 276-300.
45. World Health Organisation, A WHO report on social and biological effects on perinatal mortality. Vol 1.(Budapest, 1978).
46. Irvine Loudon, Death in childbirth. An international study of maternal care and maternal mortality 1800-1950 (Oxford, 1992) 502-507.
47. Frans van Poppel, "Religion and Health: Catholicism and Regional Mortality Differences in 19th Century Netherlands," Social History of Medicine 5 (1992): 229-253; P.A. Thornton and S. Olson, "Family contexts of fertility and infant survival in nineteenthcentury Montreal," Journal of Family History 16 (1991): 401-418; Antoinette FauveChamoux, "Les aspects culturels de la mortalite differentielle des enfants dans le passe", IUSSP, International Population Conference, Manila 1981 (Liege, 1981) Vol. 2, 341-361; M.R. O'Connell, "The Roman Catholic Tradition Since 1545," in R.L. Numbers and D.W. Amundsen eds., Caring and Curing. Health and Medicine in the Western Religious Traditions (New York, 1986) 108-145; D.T. Courtwright, "The neglect of female children and childhood sex ratios in nineteenth-century America: A review of the evidence," Journal of Family History 15 (1990): 313-323.
48. Jan de Bruijn, Geschiedenis van de abortus in Nederland (Amsterdam, 1979) 111.
49. John G. Cleland and Jerome Van Ginneken, "Maternal education and child survival in developing countries: the search for pathways of influence," in John C. Caldwell and Gigi Santow eds., Selected Readings in the cultural social and behavioral determinants of health (Canberra, 1989) 79-100.
50. Kornin,"Child maltreatment".
51. Silverman and Worden, "Children's reactions" 311.
52. Jodi Ermers, "Medea's or fallen angels? The prosecution of infanticide and stereotypes of child murderesses in the Netherlands in the 19th century," in International Commission on Historical Demography, The role of the state and public opinion in sexual attitudes and demographic behaviour (Madrid, 1990) 483-492.
53. Over et al., "The consequences".
54. Silverman and Worden, "Children's reactions", 304-305.
55. See H.P Blossfeld and G. Rohwer, Techniques of Event History Analysis (Hillsdale, N.J., 1995) for an excellent introduction to event history modelling.
56. Children of widows and widowers and those from complete families were thus included in the analysis (censored) until the moment they left Woerden. The information on date of migration was based on the population registers.
57. D.R. Cox, "Regression models and lifetables," Journal of the Royal Statistical Society, Series B, 34 (1972): 187-202.
58. G. Rohwer, TDA User's Manual (Berlin, 1997)
59. W. Rutten, `De vreselijkste aller harpijen'. Pokkenepidemieen en pokken bestrijding in Nederland in de achttiende en negentiende eeuw: een sociaal-historische en historisch-demografische studie. (Wageningen, 1997) 356-357.
60. We also studied whether the survival prospects of the child differed for various combinations of gender of the child and surviving parent. No such effect could be observed.
61. Antonia Pinnelli, "The condition of women and the health and mortality of infants and children," in N. Federici, K. Oppenheim Mason and S. Sogner eds., Women's position and demographic change (Oxford, 1993) 182-189.
62. To avoid dependence between siblings, one could also analyse first-born, second-born etc. separately (a procedure followed by Goram Brostrom, "The influence of mother's death on infant mortality: a case study in matched data survival analysis", Scandinavian Journal of Statistics 14 : 113-123) but the size of the sample is for this purpose much too small.
|Printer friendly Cite/link Email Feedback|
|Author:||Liefbroer, Aart C.|
|Publication:||Journal of Social History|
|Date:||Mar 22, 1999|
|Previous Article:||Why they stole: women in the Old Bailey, 1779-1789.|
|Next Article:||Endangered Children: Dependence, Neglect, and Abuse in American History.|