Introduction
In the last decades, promising research avenues have been opened up by interrogating the history of our most familiar and taken-for-granted statistical practices. While statistical data are nowadays routinely used in fields such as historical demography and historical sociology, the genesis of the methods and categories that define our representations of the social universe has only recently started to attract scholarly attention. The etymology of the word statistics, however, is telling in itself: statistics was originally state-istics, a scientific representation of the state for administrative, governmental purposes (e.g., Desrosières Reference Desrosières2008; Hacking Reference Hacking, Burchell, Gordon and Miller1991). Examining the history of statistical forms may thus shed light on specific structural features of modern society—if only because the establishment of power structures is necessary to gather the information or data demanded by governments and state administrations.
In terms of theory, Michel Foucault's late work has hitherto constituted an important point of departure for this approach. In Naissance de la Biopolitique, for instance, Foucault included statistical observation methods among the key biopolitical instruments. He also noted the close relationship between the elaboration of statistical methods and the rise of the idea of the “population.” Statistics or state-istics gave way to a new mode of representation. It led to the creation of both new forms of knowledge and new modes of intervention. It helped to conceive a new sort of object, society as population, which could be the target of both research and of policy interventions (see, e.g., Curtis Reference Curtis2002; Scott Reference Scott1998).
In recent decades, some historical-sociological studies have tried to trace the genesis or “genealogy” (Foucault) of distinctly modern statistical forms or categories, such as that of the inactive population or the unemployed individual (e.g., Baxandall Reference Baxandall2004; Topalov Reference Topalov1994: 312–50; Vanderstraeten Reference Vanderstraeten2006; Wobbe Reference Wobbe2012). However, the rise of the modern world also depended on forms of continuity with older traditions, on the incorporation (in one way or another) of existing forms or categories. In this paper, we will focus on the rearticulation of the household in the modern state with special regard to its state-istics. Following the Oxford Dictionary, the term has been used since the fourteenth century for a group of people (especially a family) living together as a unit; it refers to a domestic establishment (including any servants, attendants, etc.). It has often also been defined in terms of the common cooking pot: the household includes those who eat together at the same table. For the early modern era, the household is widely considered to be the most fundamental unit for social, political, economic, and reproductive purposes (see, e.g., Laslett and Wall Reference Laslett and Wall1982). In the modern era, too, the social order is often perceived to revolve around the household. Despite growing heterogeneity at the household level, this social unit remains in most discourses the proverbial cornerstone of society. In the population censuses, which have been conducted from the middle of the nineteenth century onward, for instance, household schedules have nearly always been used to gather and process individual data. But how have the expectations regarding the household as a distinct social institution shifted in the modern era? In our view, the history of social statistics may also be used to analyze the articulation or rearticulation of this notion.
A sociological history of the household, as it is depicted and defined in the world of social statistics, may not only add to the relatively small body of historical-sociological research of statistical conventions and observation methods (see Curtis Reference Curtis2002: 506; Ellickson Reference Ellickson2008: 5). It may also be relevant for the large body of historical and comparative contributions about, for example, the size and composition of the household, the cultural values that underpin its modes of solidarity, and/or the forms of management of the state's population (e.g., Laslett and Wall Reference Laslett and Wall1982; Tadmor Reference Tadmor2004; Wilk and Netting Reference Wilk, Netting, Netting, Wilk and Arnould1984). In this paper, we intend to complement the existing bodies of research by means of a case study of the Belgian population censuses. These censuses have been organized with regular, mostly ten-year intervals since the mid-nineteenth century. We will, more particularly, focus upon a period of about one hundred years: from the mid-nineteenth until the mid-twentieth century, when the welfare state more actively “took responsibility for” its population.Footnote 1
In our view, the Belgian population censuses provide an interesting case study, as it was the homo statisticus Adolphe Quetelet who played a key role in their development. Quetelet had already acquired a solid international reputation as statistician (or “social physicist”) when he established the Commission Centrale de Statistique (in 1841), tried out new procedures for census taking in the city of Brussels (in 1842), and organized the first nationwide population census in Belgium (in 1846, i.e., about a decade and a half after its independence). While Quetelet remained in charge of the censuses conducted in 1856 and 1866 in Belgium, he also defined the objectives of the Congrès International de Statistique, whose sessions were actively attended by many high-level “state servants” from countries throughout Europe and America, for more than two decades. As other researchers time and again testify, Quetelet's focus on the development of uniform methods and conventions had considerable impact on a broad range of statistical traditions both in the nineteenth century and afterward (e.g., Bracke Reference Bracke2008: 131–67; Brian Reference Brian1989, Reference Brian2002; Desrosières Reference Desrosières2008: 7–59; Porter Reference Porter1986: 41–55; Prévost and Beaud Reference Prévost and Beaud2012: 49–62; Randeraad Reference Randeraad2011). Indeed, “those who attended pushed their governments to adopt a standard template for census making on the Queteletian model” (Curtis Reference Curtis2002: 20–21). Although and because we are attentive to differences in social statistics in Europe and elsewhere, we also believe that a sociological case study of the history of Belgian statistics and statistical constructs may have broader relevance.
Our primary historical sources consist of the Belgian census reports. These reports contain a presentation of the results of the census, as well as a rather technical part that reproduces the instructions to the census takers and local authorities.Footnote 2 We will particularly focus on the ways in which data about the household were collected and processed in the successive Belgian censuses. While the census questionnaires had to be intelligible to both the census takers asking them and the population answering them, the changes in the descriptions and definitions of the household may reflect more fundamental shifts in our ways of representing society and conceiving of the population. We therefore believe that the world of statistics and state-istics allows us to develop a sociological history of the household in the modern era.
Hereafter, we start with an overview of the development of the enumeration techniques used by the Belgian statisticians. We then consider changes in the representation of both family households and collective households, such as barracks, religious communities, hospitals, or prisons. Next, we discuss the changing expectations with regard to the membership of these households, to the inclusion and exclusion of particular individuals as household members. In the concluding section, we formulate and discuss some more general conclusions regarding the “reorganization” of the modern population and its basic social units.
Enumeration Forms
Although a population census is mostly defined as an “attempt to count all the people in a country at a given point in time” (Headrick Reference Headrick2000: 76), social statisticians have often distinguished between two definitions of the population. The total population may comprise either all “usual” residents of the country or all persons present in the country at the time of the census. The total of all usual residents is referred to as the de jure population and the total of all persons present as the de facto population. In Belgium, both principles have been used side by side, although the statisticians’ emphasis shifted in the latter half of the nineteenth century from the de facto to the de jure principle. In this section, we will trace the adaptations of the census questionnaires and enumeration forms, while the technical “improvements” of the statistical armamentarium also embody structural changes in the idea of the population and of the household as its basic social unit.
In Belgium, the 1846 population census was mainly based on the de facto principle, which was considered to be the simplest and most efficient method at that time (e.g., B 1846: X; B 1856: VIII, XLV, LIV, LV, LXXVI; see also Thorvaldsen Reference Thorvaldsen2006: 84–85). The census takers made use of household schedules, on which all persons living within the household had to be listed. Two types of schedules were used: one for “typical” family households and one for institutions or “collective bodies” (corps collectifs), such as barracks, monasteries, or prisons. In principle, the head of the household or of the institution had to fill out the questionnaire. If the head was absent or unable to complete the form (e.g., because the head was illiterate), other household members or the census takers had to fill out the form (B 1846: L, LXIX). According to the official instructions, the census takers had to control and, if necessary, correct the information provided on each questionnaire.Footnote 3
The household schedule collected data about a wide variety of issues, such as age, gender, place of birth, language, religion, civil estate, occupation, reading and writing skills, and reliance on public support (see, e.g., B 1846: X, XXV). In line with the de facto principle, information about the whereabouts of all individuals present in the household or institution had to be provided. But the 1846 census also distinguished between three types of residences: habitual, temporary, and de passage. Hotels or dwellings, where people stayed for less than one month, were considered to constitute résidences de passage. Institutions that sheltered individuals for more than one month, such as prisons, barracks, or boarding schools, were called temporary residences. Generally, however, individuals were expected to be in their usual residence on census day (e.g., B 1846: L). For the statisticians, the distinction between these types of residences made it possible to identify and count the so-called floating population (as part of the de facto population). In 1846, the individuals also had to be listed in a specific, prescribed order on the household form: the head of household, the spouse, the children, the in-living family or nonfamily members, the in-living servants, and finally the individuals who were but temporarily present in the household. Although the main concern was with the de facto population, this last category thus again served to distinguish between the floating population and the habitual, resident population.
In 1856 the individuals had to be listed on the household schedule in the same order as in 1846, although it was added that the children had to be ranked by age. But the household schedule now also included a field in which information on the habitual residence of temporary visitors had to be listed (B 1856: XXXVIII–XXXIX, XLI–XLII). In addition, the head of household was explicitly instructed to provide the information for individuals who did usually belong to the household but who were temporarily absent. The 1856 census thus also included and counted absentees, at least if they were expected to return to their habitual place of residence (B 1856: XLVIII, LVI). This may be regarded as the first deliberate attempt to apply the de jure principle, as well as to distinguish between the de facto and de jure population.
In the 1866 census, the de facto and de jure method were officially used side by side in Belgium. On the household schedule, the list of individuals was to be divided into two separate groups: those who belonged to the household, whether present or temporarily absent on census day, on the one hand, and those who were present in the household but did not belong to it, on the other (e.g., B 1866: VII–VIII, XI–XII, XXXVI, XLII). A field was also added to provide information about their actual residence (séjour réel) and their habitual or regular place of residence (résidence habituelle) on census day. To guard against overenumeration, the Belgian statisticians started to reiterate that individuals could only be a member of one household or institution at the time (e.g., B 1866: XII). For each individual in the household, the degree of parenthood with the head of household had to be indicated, too (B 1866: IX, XXXII–XXXIII).
Following a recommendation made by the International Statistical Congress, the Belgian statistical authorities started to conduct censuses in years ending in zero (Quetelet Reference Quetelet1873: 121). Therefore, the fourth population census, which was originally scheduled for 1876, ended up taking place in 1880.Footnote 4 In the 1880 census, two extra census forms were distributed (see, e.g., B 1880: IX). Next to the household schedule, special forms were introduced for the individuals who were only temporarily present in a dwelling on census day. A distinction was made between the individual card (carte individuelle) for the “strangers” or temporary visitors in family households and the bulletin special, which was used to collect the same kind of information for all the “inmates,” who were expected to reside only for a limited time in the collective bodies. In 1890, these forms were reused but named bulletin spécial personnel, on the one hand, and bulletin spécial collectif, on the other (B 1890: IV, CXXXI). As such, it proved to be a relatively stable statistical innovation; together with the household schedule, both bulletins remained in use throughout most of the twentieth century (see B 1900: CVI–CIX; B 1910: 44–47; B 1920: 18–21; B 1930: 46–51; B 1947: 108–13; B 1961: 49–51; B 1970: 46–49). As a consequence, the “visitors” were immediately set apart and counted by means of separate forms. By focusing on the number of individuals formally belonging to the household in its regular place of residence, the Belgian censuses also began to elevate the de jure principle above the de facto one (see, e.g., B 1880: LXXXVIII; B 1890: CXXXI; B 1900: X; B 1910: VI).
In the discourse of the statisticians, these “technical” changes in/of the enumeration forms mostly constituted innovations and improvements. They were accompanied by numerous instructions to the census takers and heads of households, who had to complete the forms. However, these technical improvements also reveal basic changes in the idea of population. The statisticians developed a particular idea at the cost of others. In contrast with the de facto method that informed about the “real presence of individuals” (présence réelle), the de jure method focused on individuals’ official or regular residence (B 1890: XIV). The de jure counts emphasized the place where individuals ought to be on census day, regardless of the place where they were actually staying on that day. Concomitantly, the criteria for household membership shifted. The first, de facto censuses registered all individuals who were staying at the same place on census day as members of the same household. From the 1880 census onward, a regular residence became a sine qua non for household membership. Only the habitual or regular residents could be registered on a household sheet; only de jure habitants could be counted as household members (e.g., B 1880: IX, CXXII; B 1890: CLIV, CLXXVIII; B 1900: IV; B 1910: 2; B 1920: 3; B 1930: 5). The state thus became involved in (re-)defining the household and the conditions for individual membership of the household and the population. As we will see, the history of the enumeration principles and practices not only—mostly indirectly—informs us about the difficulties that this legalization process brought about, but also displays the practical and state-istical consequences of this shift.
Family Households
As mentioned before, the household is the basic population unit for the censuses. In the first Belgian censuses, the household was described as “the union” (la reunion) of two or more individuals who were “living together” (vivant en commun) (B 1846: L, LXIX; B 1856: XXXVIII, LIV). But after the adoption of the de jure method, the statisticians were forced to be more specific about the criteria for household membership. From 1880 onward, they mostly defined the household in terms of habitual co-residence. Although the rather vague description l'ensemble des personnes was used in the 1880 and 1890 censuses (B 1880: XLV; B 1890: IV), the statisticians typically made use of phrases like “habitually living under the same roof” (résider habituellement dans la même maison/habitation) or “sharing a common life” (avoir une vie commune) (e.g., B 1880: XLV; B 1890: IV; B 1900: LXXXII; B 1910: 28; B 1930: 6). In its specific historical details, however, the criteria for household membership also displayed some important implications of state-istical decision making.
From the first census onward, it was already specified that “singles” (les individus de l'un ou de l'autre sexe, vivant seuls) could be considered to constitute a household of their own (e.g., B 1846: LI; B 1866: XXXIX). From the 1900 census onward, the statisticians distinguished more explicitly between a “simple unit,” which consisted of “one person living alone” (une personne vivant seule), and a “collective unit,” which was formed by “the union of two or more persons” (la réunion de deux ou plusieurs personnes) (e.g., B 1900: LXXXII; B 1910: 2–3, 28; B 1930: 6).
Of more conceptual concern was the relation and distinction between the “household” and the “family.” In the first censuses, these terms were occasionally used as synonyms (B 1846: XLV, L; B 1856: LX; B 1866: XI). At the same time, however, attention was directed to the distinction between both social institutions. For instance, it was stipulated in the instructions to the census takers that live-in servants or workers had to be treated as members of the households in which they were employed (e.g., B 1846: L; B 1856: LIV, LXXII; B 1866: VIII, LXXV, LXXXV).Footnote 5 In the 1846 and 1856 censuses, the information about the heads of households and their relatives also had to be listed prior to that of the other residents on the household schedules. From the 1866 and 1880 censuses onward, the census takers were more explicitly instructed about the distinction between co-residential groupings and family or kinship structures (e.g., B 1866: XXXIX; B 1880: XI; B 1890: CXVI; B 1900: LXXXIII; B 1910: 2–3, 28; B 1920: 4; B 1930: 6). It now became routinely reiterated that households were not characterized by kinship relations (see also B 1880: XLV; B 1890: IV). But such instructions also defined “the facts,” which the censuses had to inform about.
Initially, the elaboration of the distinction between household and family mainly provided the background for analyses of family and kinship relations. The 1866 census started to inquire into the family ties established by marriages. In 1846 and 1856 only an indication of marital status was required (with two options: married or widowed), but the 1866 questionnaire also asked for the name of people's (former) spouse (B 1866: XLI). After 1866, it was stressed that married women could not use their own family name, but had to be registered under their husband's name—even when they were no longer married, but widowed or divorced (e.g., B 1890: CXX, CXXVI; B 1900: CIV–CIX, CLIII).Footnote 6 These women thus became defined by their (former) marriage; men were expected to represent the family households. In this sense, the censuses both confirmed and reinforced the subordinate position of women in the family household (see also Wobbe Reference Wobbe2012; Zimmermann Reference Zimmermann2001).
Moreover, the 1866 census started to inquire explicitly into the filiation between the head of household and the other residents in the household. “Popular terminology,” instead of a formal description of the degree of parenthood, was allowed in order to clarify this filiation (B 1866: XLI). Among the examples provided were terms such as son, grandson, son-in-law, nephew, cousin, and uncle. According to the summary report, this part of the census questionnaire was specifically intended to gather information about “the filiation within households that bring together several generations of the same family” (en établissant la filiation dans les ménages qui réunissent plusieurs générations de la même famille) (B 1866: IX). The wording of the item and the related instructions to the census takers also shows that the lineage with the pater familias was considered to be a defining characteristic of each family member.
Similar items and instructions were used in the following population censuses (see B 1880: XLVIII–XLIX; B 1890: LXXXII–LXXXIII, CXX). But, after that the distinction between household and family had been firmly established, more response options for this item were allowed for. From 1900 onward, residents without kinship relation with the head of household also had to inform about their “position” or “situation” (position, situation) in the household and their relation to the head of household. Among the examples provided in the instructions were descriptions such as governess, tutor, servant, pupil, friend, and associate. In cases that the relationship could not be specified, the answer “none” was allowed (B 1900: CLIV; B 1910: 2–3, 103; B 1920: 10; B 1930: 6; B 1947: 104; B 1961: 43). Like the family, the household was depicted as a unitas multiplex, as a structured whole (see also Luhmann Reference Luhmann1990: 196–217). Perhaps this extension of response options first of all signals a growing awareness of the household as a co-residential voluntary association that may bring together different individuals in different constellations.
In sum: the relative neglect of the distinction between household and family in the mid-nineteenth-century censuses may be seen to still reflect the traditional, lasting importance of the family household as the basic social, co-residential unit of individuals who share in common consumption and production. The emphasis on the difference between family and household, as it can be observed in the censuses of the latter part of the nineteenth century, signals the institutionalization of different forms of understanding. In the last decades of the nineteenth century, the household could no longer be depicted as a traditional and “natural” reality. Gradually it became perceived as a voluntary association, even if the statisticians initially focused most of their attention on the natural, kinship relations within the family (as part of the household). In the twentieth-century population censuses, the family household became analyzed as a dynamic—relatively unstable—institution, to which individual members might belong for various reasons. The social statistics rely on this Gestaltswitch, although the “printed numbers” do not shed light on it.
Collective Households
Because of its focus on co-residential groupings, the term household could also be applied to individuals living together in “collective bodies” (corps collectifs), such as monasteries, nursing homes, asylums, prisons, workhouses (providing employment for paupers and support for the infirm), barracks, orphanages, and so on.Footnote 7 As mentioned before, two types of schedules were used in the first Belgian censuses: one for the typical family households and one for the collective bodies (B 1846: XIV; B 1856: XXXVIII, XXXIX).
The extension of the term household was, however, not obvious. In fact, the term was initially only used for the typical family households (B 1846: L; B 1856: XXXIX). In 1856 and 1866, it was added that each type of collective body was “assimilated to a household” (assimilé à un ménage) (B 1856: XXXIX; B 1866: XXXIX).Footnote 8 From the last decades of the nineteenth century onward, (several of) these collective bodies were no longer just “assimilated to,” but presented as “real” alternatives to households (e.g., B 1880: X; B 1890: IV, LXX). In the censuses of 1900 and 1920, for example, both family and collective households were described as elementary social groups (groupe sociale élémentaire) (B 1900: XXVI) or elementary social collectivities (collectivités sociales élémentaires) (B 1920: 4). Such statistical formulations may reflect the growing social importance of “total institutions” in the modern era. On the basis of their household definitions, the “disciplinary institutions” may be dealt with in the same way as the “typical” family households (see also Foucault Reference Foucault1978: 107; 2006: 47). Interestingly, however, there were also differences in the way in which the meaning of the term household was extended by the Belgian statisticians.
Special attention was given to religious communities, such as orders and congregations (see Vanderstraeten Reference Vanderstraeten2014). But despite the strong tensions between state and church in nineteenth- and twentieth-century Belgium, the statistical definition of religious communities did not cause many problems. For the statisticians, membership of these organizations was clearly regulated. Their members lived in splendid isolation behind high brick walls; they had taken religious vows (chastity, obedience, and poverty) and committed to spending their entire life in the religious organization. It was also commonly accepted that all members had joined voluntarily, and that the convent constituted their place of residence (e.g., B 1880: X). In this light, it seemed obvious to treat these total institutions as households, characterized by a joint, shared life (e.g., B 1900: V, LXXXIII).
But more ambiguity existed regarding the “status” of a number of other total institutions, such as prisons, barracks, hospitals, sanatoria, madhouses, hospices, homes for the elderly, or boarding schools. Census after census, the statisticians tried to draw a distinction between institutions that either temporarily or permanently sheltered individuals, such as sanatoria and hospitals, on the one hand, and residences for the elderly and incurables, on the other (e.g., B 1900: LXXXIII; see also B 1910: 2, 28; B 1920: 18–19, 109; B 1930: 6–7). Time and again, the census takers were instructed to deal with institutions of the first type as temporary, but not as habitual places of residence for their inmates. By consequence, the “inmates” were perceived as formally belonging to another family household, to which they were believed to return at the end of their posting, after having served their sentences, after having been treated for their illnesses, and so forth. In the population census, the inmates thus had to be reported as being momentarily or temporarily absent from their last residence. Despite the fact that they could be assigned for many years (if not for life) to these institutions, their habitual place of residence had to be somewhere else. According to the instructions, these inmates had to be enumerated “back home” in what was considered to be their habitual place of residence (e.g., B 1900: IV; B 1910: 27–29; B 1920: 37, 109; B 1930: 27, 158).
In other words, the census takers had to reassign these individuals administratively to places considered more appropriate. In practice, however, this requirement often proved difficult to fulfill. Time and again, exceptions had to be granted. In the 1866 census, for example, it was noted that institutions, such as prisons and colonies, could be the habitual place of residence for “the detained and the enclosed who have no parents, no residence, and who do not know where to settle after their release” (des détenus et des reclus qui n'ont ni parents, ni habitation, et qui ignorent où ils se fixeront à leur sortie) (B 1866: XLI). As of 1890, a more general formulation was used to indicate the special circumstances that could provide the ground for exceptions: the institution could constitute the habitual place of residence for all the inmates who did not belong to any other household (B 1890: CXVII; see also B 1900: CXLV; B 1910: 96; B 1920: 109; B 1930: 7). Only in exceptional cases could the disciplinary institutions constitute households in themselves. Or stated differently: the statisticians remained cautious in extending the term household to disciplinary contexts. Despite the fact that their definition allowed and forced them to apply the term household to all co-residential groupings, they continued to use other, mostly unarticulated criteria to redefine the idea of household and household membership. Their instructions reveal a hidden preference for a social order based upon family households.
A Place of Belonging?
Co-residence is the main distinguishing feature of the household. Obviously, however, co-residence has a social dimension, too. In this section, we analyze whether the household was also regarded as a social and emotional setting that embodied a sense of belonging.
Not coincidentally, the statisticians’ language was often formal. The household was said to be “formed” (formé) (B 1846: LI; B 1866: XXXIX; B 1880: XI; B 1890: LXX; B 1900: XXIV; B 1910: 3; B 1930: 6), “composed” (composé) (B 1846: XXIX; B 1856: LXXXII; B 1890: LXVIII; B 1900: LXXXV; B 1910: 3; B 1920: 10; B 1930: 29), or “constituted” (constitué) (B 1880: XI; B 1890: IV; B 1900: CXLV; B 1910: 95; B 1930: 26). The individuals within the household were considered to be its constitutive parts (B 1846: L; B 1856: LXXXII; B 1866: LXXXIV; B 1880: X; B 1890: LXVIII; B 1900: XCV; B 1910: 2; B 1920: 10; B 1930: 5).
More “colored” expressions of belongingness, however, also appeared. In 1846, for instance, the term membership was not yet used in the context of the household. While persons were included (comprises) in the household, only the family had “members” (e.g., B 1846: LII, LIV). But from 1856 onward, the statisticians also started to speak of household membership (e.g., B 1856: LXXXII; B 1866: VIII; B 1880: X; B 1890: LXVIII; B 1900: CLIV; B 1910: 6; B 1920: 29; B 1930: 10). A similar shift in the use of the term belonging (appartenir) took place. Initially, this term was reserved for the family and its members (B 1856: LXXII). However, its use was extended to the household and household members as of the 1866 census (B 1866: XL; B 1880: XLVIII; B 1890: CXXXIII; B 1900: CXCV; B 1910: 2; B 1930: 6). As we have already pointed out, the state-istical interest in intrahousehold relations gradually also transcended the level of family or kinship relations. From the 1900 census onward, the statisticians adapted the questionnaire items dealing with intrahousehold relations to include nonfamily or nonkinship relations in the household, too. In other words: for the statisticians, the household gradually developed into a social entity sui generis, evoking a sense of belonging. Interestingly, this shift not only applied to the “typical” households. We found similar developments for the collective households. They were initially described as corps collectifs, as organized social bodies. In the early twentieth century, the statisticians occasionally also underlined the existence of some forms of “solidarity” (as Durkheim would say) within all households (e.g., B 1900: XXVI; B 1920: 4).
From 1866 onward, the questionnaires had to inform about the nature of the bonds between the household members and the head of household—not about the bonds of all the household members among each other. Perhaps mainly for reasons of statistical calculability, the household members were listed primarily as dependents of the head of household; they were registered as, for example, “the son of the head of household,” “the spouse of the head of household,” or “the mother of the head of household” (e.g., B 1890: CXX). As we mentioned before, the procedures for the registration of family names also confirmed and reinforced the central position of the head of household. In the statistical representations, the relational network within the household thus was expected to revolve around the head of household. In the prototypical case (i.e. for the family households), “belongingness to the household” was translated in terms of “dependence from the male head of household.”
At the start of the twentieth century, at the moment that an inventory of nonnatural bonds was demanded, other forms of depicting the household structures could be probed. But the Belgian statisticians remained inclined to focus on hierarchical structures. This is perhaps best illustrated by two technical “innovations,” with which they experimented in the first half of the twentieth century. In the 1920 census, the Belgian statisticians introduced a special individual card for each head of household in addition to the household sheet. This special card was used to gather more detailed information about these heads and their households and to separate this information from that of the “mass” of other data (B 1920: 3–4). A quite similar special individual card was used in 1930 in order to acquire more detailed information about all married men and their offspring (B 1930: 10–11). Both innovations did not have a lasting effect; these special cards did not reappear in the censuses conducted after World War II (see B 1947: 51–52). But both these innovations make clear that the Belgian statisticians continued to depict the household and the family in hierarchical terms in the early twentieth century. Heads of households and married men were singled out; the other members of the family or household were regarded and classified as dependents.
As we have already seen, the statisticians also began to reiterate the rule that every individual could only be a member of one household at a time in the latter part of the nineteenth century (e.g., B 1866: XII; B 1900: LXXXII). Technically, this rule had to be emphasized in order to guard against overenumeration in the de jure population censuses. Remarkably, however, the notion of “not belonging” to the household also emerged as a complement of the legalization of household membership and of the conceptualization of the household as a place of belonging. Its diverse manifestations have already been discussed. From the 1866 census onward, those present in the household on census day had to be divided into two separate groups: the members of the household and the individuals who were regarded as “being foreign to the household” (les individus étrangers au ménage) (e.g., B 1866: XL; B 1900: XCIV; B 1910: 39; B 1930: 30).Footnote 9 Especially with regard to the residents of collective bodies, it then became common practice among the census takers and statisticians to focus on household membership status. In most cases, they had to reassign these residents to other family households. As we have seen, the variety of instructions and regulations dealing with the population of the “total institutions” also incorporate normative expectations underlying the statistical notion of solidarity within, or belongingness to the modern household. Not all forms of “living under the same roof” were considered to be of equivalent social value. The census takers repeatedly had to distinguish between “normal” and “abnormal” forms of belongingness.
In the Belgian censuses, individuals have always been registered on household sheets as household members. But the meaning of “household” and “household membership” shifted substantially in the era between the mid-nineteenth and mid-twentieth century. The distinctions that were drawn and the relations that were depicted in the households also shifted considerably in this period. The changing statistical perspectives on belongingness illuminate the shifting concerns of the state with its population; they shed light on the premises on the basis of which the welfare state developed alternatives to kinship and household dependence in the era after World War II.
Conclusion
Although the history of social science has usually been written with an emphasis on social theory, social science was from the beginning also a project of collecting and analyzing facts or data. Much of this work was performed by state administrations. It was also bound up with systematic “governmental” interventions. In this perspective, a basic task of social science in general and of state-istics in particular was to work out forms of observation and classification appropriate to the new, “modern” social conditions (Porter Reference Porter, Daston and Lunbeck2012: 299). At present, it can also be maintained that statistics has become a way of “making up people” (Hacking Reference Hacking, Burchell, Gordon and Miller1991). It has become a way to understand, interpret, and reform social reality.
Our interest in this paper has not been in the observed facts or the data as such, but in what interested the official observer, the state-istician. We have focused on the administrative tools, which were constructed and reconstructed by the statisticians. In particular, we have focused on the variability of the concepts and definitions of the household, which have been used in the Belgian population censuses, and which constitute in our view traces of the intentions and conflicts that presided over the production of these data.
As our findings show, the mid-nineteenth-century statistical notions of the household were still closely linked with early modern ideas and ideals. The household was perceived to be a natural, “sovereign” institution grounded in kinship. The articulation of the household in de jure terms, which took place in the latter part of the nineteenth century, reflected and enhanced the trend toward a “governmentalization” of this social institution. The state not only became entitled to control and regulate membership within this institution, but also its state-istics contributed to specifying “normal” household structures or relationships.
Of course, statistical definitions of the household had to come to terms with changing social realities, changing forms of co-residence in modern society, and new expectations regarding the role of individuals in this co-residential unit. But the process of defining and redefining the household was also one of endorsing particular ideas and ideals about the household. The state-istical and de jure view on the household also mobilized specific norms or evaluative standards. It disciplined and ordered the reality it aimed to measure. The observation and classification forms of statistics not only identified the cornerstones of society, but also depicted the “normal” household in very specific hierarchical terms. The Belgian censuses not only focused on the de jure population members, but also reinvigorated a view of the social order based on family households and their male heads. They not only distinguished between those who belong and those who do not, but also articulated specific expectations regarding solidarity within the household and the population. As such, statistics thus certainly do not merely collect scientific facts about the state of the state. The preceding “genealogical” analyses have shown how the population statistics also served to produce and reproduce “normalized” individuals and households. L'homme moyen (Quetelet) also is a result of the conceptual and administrative organization of statistics.
Appendix
Belgian Census Reports
B 1846Population. Recensement général (15 Octobre 1846). Bruxelles: Le Ministre de l'Intérieur, 1849.
B 1856Population. Recensement général (31 Décembre 1856). Bruxelles: Le Ministre de l'Intérieur, 1861.
B 1866Population. Recensement général (31 Décembre 1866). Bruxelles: Le Ministre de l'Intérieur, 1870.
B 1880Population. Recensement général (31 Décembre 1880). Bruxelles: Le Ministre de l'Intérieur, 1884.
B 1890Population. Recensement général de 1890. Tome I. Bruxelles: Le Ministre de l'Intérieur et de l'Instruction Publique, 1893.
B 1900Population. Recensement général du 31 Décembre 1900. Tome I. Bruxelles: Le Ministre de l'Intérieur et de l'Instruction Publique, 1903.
B 1910Population. Recensement général de la population du 31 Décembre 1910. Tome I. Bruxelles: Le Ministre de l'Intérieur, 1916.
B 1920Population. Recensement général du 31 Décembre 1920. Tome I. Bruxelles: Institut National de Statistique, 1926.
B 1930Bevolking. Algemeene telling op 31 December 1930. Ie Boekdeel. Brussel: Nationaal Instituut voor de Statistiek, 1934.
B 1947Algemene Volks-, Nijverheids- en Handelstelling op 31 December 1947. Deel 1. Brussel: Nationaal Instituut voor de Statistiek, 1949.
B 1961Algemene Volkstelling, Woningtelling en Handels- en Nijverheidstelling op 31 December 1961. Onderrichtingen aan de Gemeentebesturen en Handleiding van de teller. Brussel: Nationaal Instituut voor de Statistiek, n.d.
B 1970Algemene Volkstelling, Woningtelling en Handels- en Nijverheidstelling op 31 December 1970. Onderrichtingen aan de Gemeentebesturen en Handleiding van de teller. Brussel: Nationaal Instituut voor de Statistiek, n.d.
