'I Is; Therefore I Am': The Census as Practice of Double Identification

by Evelyn S. Ruppert
The Open University

Sociological Research Online 13(4)6
<http://www.socresonline.org.uk/13/4/6.html>
doi:10.5153/sro.1778

Received: 28 Mar 2008     Accepted: 18 Jul 2008    Published: 31 Jul 2008

Abstract

I examine practices of modern census making with a specific focus on Canadian censuses of population from 1911-1951. My analysis builds on the work of two recent and related streams of research in the social sciences. One draws from Foucault's writings on biopower and post-Foucauldian governmentality studies. It examines the census as a political technology that produces a specific knowledge or political arithmetic (statistics) of the population, so that its forces and strengths can be acted upon by various state authorities. The census is thus understood as a field for the administration of the state. The other focuses on how censuses are socially constructed, on the 'making' of censuses as opposed to the 'taking' of censuses and the use of census data as 'evidence'. These studies document how the interests and political influence of various actors shape census making. The census is thus understood as a particular way of defining, collecting and organising social observations about individuals and not a simple reflection of an empirically existing reality. While the two streams of research have usefully challenged the facticity of census data, they have tended to reinforce a division between the real and the constructed. For if census data is not 'real' but a particular construction then what exactly does it represent? I contend that censuses are part of myriad identification practices that have come to produce subjects who are able to recognise and identify themselves in relation to the categories constructed and circulated by the census. It is through processes of double identification (state-citizen) that census categories come into existence, become facts and can then in turn not only be measured, analysed and assembled (objectification) but also be identified with (subjectification). The presence of such double identification makes an ostensible division between facticity and representation artificial.

Keywords: Censuses; Social Constructivism; Realism; Objectification; Subjectification; Canadian; Governmentality; Identification

Introduction

1.1 Numbers occupy a privileged status in modern political culture. Discourse and debates in the media, academia and government on everything from the economy and health to crime and immigration are often driven and framed in reference to numbers. From the apportioning of political representation and the assessment of a government's performance to the formulation of government programs and policies, numbers are often deployed in these various arenas as objective criteria to evaluate, judge and govern (Rose 1991). In fact it is now hard to conceive of governing without numbers on voting, opinions, sales, taxation, gross national product, births, deaths, migration, crime, divorce and so on. The significance of numbers is no better evident than in the making of population statistics through the census, one of the oldest and most regularly used techniques. Through censuses most modern states have constructed an infrastructure of numbers that serves not only as a source of state legitimacy but also shapes the regulatory capacities of state authorities themselves. Indeed, most state departments such as economic development, health and criminal justice depend on population numbers constructed from censuses. And finally, as members of a population we often assess our relative status and life situations in relation to census statistics on health, housing, living arrangements, conjugal status, income, ethnicity and so on.

1.2 It is through the identification of individuals with numerous categories that census numbers and statistics are constructed. Censuses do not simply involve 'counting heads', but are part of a broader configuration of historical practices involved in the constitution and classification of identities as legal and bureaucratic categories that have been fundamental to the multiple operations of the state (Caplan and Torpey 2001). It is through a variety of practices of identification that we know a population or become a 'legible people' (Scott 1998). The census is one of the few administrative practices concerned with knowing population that is repeated, albeit with varying degrees of consistency, by most western states every five or ten years. It has become both relatively taken for granted as well as a deeply embedded cultural practice of documenting identification and governing population by numbers. It is part of a broader set of information collecting practices that arguably constitute one of the emblematic and greatest achievements of the modern state (Higgs 2004b). Since the nineteenth century the practice has been closely tied to the international statistical movement, which has sought to standardise and normalise national practices and the making of population across time and space (Goyer and Domschke 1992). Indeed, practices in the nineteenth century, a period described by Hacking as the 'avalanche of numbers', were tied into a trans-Atlantic network of intellectuals, politicians and civil servants that was fuelled by state competition in the race for progress (Curtis 2001). Over the past century, the practice has been institutionalised, codified and systematised such that many policies and practices of governments, international organisations, corporations and researchers rely upon censuses to a great extent. That in most cases every inhabitant of a state is legally compelled to participate in the practice is an additional unique and intriguing characteristic that attests to its embeddedness as a modern cultural practice of governing by numbers.

1.3 This investment in numbers has been durable despite several decades of scholarship that has challenged the consistency and facticity of census data. Social constructionists have argued that once reported censuses conceal processes of construction such as the political interests and aspirations of various groups or social actors that influence census officials, and both the questions and acceptable answers (e.g. Curtis 2001; Iacovetta and Mitchinson 1998; Kertzer and Arel 2002). They have asserted that censuses are more representation than reality. However, scholars have by and large remained committed to the data value of censuses and have continued to focus on debates about their accuracy towards better grasping empirical questions. Thus there are innumerable studies that use census data but few studies about how the data is constructed. Political and administrative uses of the census are also well studied and documented as are technical disputes about the accuracy of data and the classification and measurement of particular population characteristics. The accounts do not deny political influence and the engagement of numerous actors, but when acknowledging this there is a tendency to be narrowly realist: that through better construction and through the inclusion of different interests and voices, censuses will better approximate the 'real' population. And finally, as the standard guide to international censuses indicates, a realist interpretation dominates: the census is a 'reflection of society', a 'photograph of a population at one moment in time' (Goyer and Domschke 1992).

1.4 I do not propose to engage in this debate but rather develop an alternative interpretation of censuses and the numbers they construct. While agreeing that the population is something that is constructed rather than discovered and made possible through objectifying techniques such as establishing conventions of equivalence between bodies, what has generally been overlooked is how we have come to identify ourselves as members of a population through the census and what capacities and agencies were necessary for such identification to be possible. That is, the census produces more than simply numbers and statistical objects. I investigate the census as a reflexive practice of identifying oneself in relation to a whole, a practice that involves recognising oneself as a member and part of a population that is connected to myriad practices involved in the formation of a political subjectivity. This aspect of the census, of liberal subjectivity and producing population, constitutes census taking, which is when the capacity to identify with the population is cultivated and subjects of the census are brought into being.

1.5 Taking into account both understandings, I interpret the census as a practice of governing that integrates totalising and individualising procedures of governing. This is an aspect of what Foucault calls a political technology of individuals through which 'we have been led to recognise ourselves as a society, as a part of a social entity, as a part of a nation or of a state' (Foucault 1994: 404). Totalising procedures are the bio-political technologies (knowledge) that make it possible to know the nature and then govern and regulate the forces of the body, that is, the population. To govern this entity a specific knowledge is required to construct the census object. But the site through which the population is accessed and governed is the individual body. Individualising procedures consist of both anatomo-political technologies (discipline) that shape the conduct of the individual body as well as subjectifying techniques that cultivate the capacities of reflection and self-discipline. The census is a technique that integrates both individualising and totalising technologies: through it the object of population is constructed and regulated and the subject of population constituted. Below I interpret this as 'double identification': through the census the subject identifies herself as part of the population and the state identifies the subject and assembles the population. In so doing, I am opening up questions about the link between state practices and the formation of identification categories, collectivities and subjectivities more generally.

1.6 I undertake this by returning to the first few decades of the twentieth century when census taking in Canada was in the early stages of becoming a key technique of governing the state. At that time, the census signified a new interaction between the state and the individual. It involved an at-the-door interview with an enumerator and required that the individual identify herself and other members of her household in relation to the state in a new way. It inculcated a particular way of thinking of the relationship between the individual and a larger social entity - the population - a way of thinking that is now relatively taken for granted.[1] I suggest that census taking produced population one subject at a time whereby subjects gradually, but fitfully, acquired the capacity to recognise themselves as members and parts of a whole. Producing population captures the practice of developing, creating and bringing this capacity into being, whereas constructing population attends to the practice of assembling and representing population (census making). The aim of this paper is to understand the former, how census taking is part of myriad state practices that have been part of the formation of a political subjectivity: the reflexive capacity to recognise and identify oneself as a member of a population.

1.7 The understanding that census taking produces population arose out of my encounter with original census manuscripts on the taking of censuses in early 20th century Canada. For the past five years I have been engaged in a project (see < http://www.canada.uottawa.ca/ccri/CCRI/index.htm >) involving the construction of a series of quantitative and qualitative databases related to the 1911-1951 censuses of Canada.[2] My initial reason for viewing original census manuscripts was to contribute to the design of rules and protocols for extracting a sample of 'microdata' or 'individual level data' by translating the handwritten material into a numeric and digital format. Hitherto only the 'macro' has been visible, that is, the translation of the census manuscripts into coded strings of data and aggregate statistics produced by the state's census bureau.

1.8 By returning to original census manuscript forms, microdata would reveal the variability of lives that is lost in aggregate statistics and provide insights about the lives of the historically anonymous, that is, all those individuals who do not leave records of their lives. Thus, the objective was to 'see' individuals instead of population aggregates and to produce 'new' quantitative datasets that would provide a new evidentiary foundation for the study of individual lives and the making of modern Canada. Additionally, by assigning a geocode to each individual spatial analyses of social change would also be made possible.[3] As the databases would span five decades and be harmonised with those compiled or being compiled for 1871-1901 and 1961-2001, it would be possible to undertake longitudinal analyses and study social change from the vantage point of individual lives rather than from the aggregate.

1.9 However, after working with thousands of completed manuscript forms, I began to think about the paper technology of the form itself and its role, along with that of the enumerator in mediating the relation between the state and the individual. The census manuscript form in the early twentieth century consists of a grid of columns covering each question or classification of the population (name, address, age, sex, marital status, ethnicity, etc.) and of rows for categorising individuals in relation to each classification or question (e.g. male, female, single, married) and their social relation to others within a household (e.g. wife, son, niece). See an example here at Figure 1. It is quite different from the standard and familiar self-administered questionnaires used today, which generally consist of a list of questions and possible responses.

1.10 By contrast, in the early twentieth century, the form is a blank grid that represents a social space within which each individual is to be identified. It also designates a spatial hierarchy of households, dwellings, buildings, streets, towns, cities, townships, and counties within which each individual is to be located. These relational spaces were lost after the census bureau translated the text on the forms into numbers and tables and relegated the original manuscripts to the confidential files of the state. Returning to the original manuscript form reveals this relational space, a structuring template that circulates the state's administrative classification and objectification of the population. Significantly, categories do not appear on the form, however, enumerator instructions provide examples of expected and accepted answers. In addition to the paper form, enumerators are thus key actors who mediate the interaction between the state and the individual. The mediation is evident in the result: the completed manuscripts reveal much variety and uncertainty rather than straightforward translations of each individual into the state's classification system. It is a variability that is particularly evident during the construction of microdata when the numbers of unique answers to individual questions are more numerous than the categories reported in the aggregate statistics. For instance, a four percent sample of individuals enumerated in 1921 (approximately 362,041 individuals) resulted in over 563 unique versions of 'racial and tribal origins'. However, the aggregate statistics identify 32 unique categories and five 'other' categories. Beyond the obvious spelling errors and problems of illegible handwriting, how can we account for the number of unique versions?

1.11 One account is to treat the manuscript form as a paper technology that seeks to direct and shape messy, imprecise and individual narratives into a form that can be added up to something called population. It is a particular way of structuring and organising the way that individuals narrate their identification with the population. Because it is a blank form, its completion depends on the enumerator acting as a mediator between the individual narrative and the preformatted grid of the manuscript. Indeed the role is evident in the completed form, which is a meticulously handwritten document where details about individuals are painstakingly recorded, often in immaculate handwriting. (See Figures 2, 3 and 4). The imprecision of the individual narratives and the difficulties of mediation are sometimes to be found in the notes in the margins, erasures, crossed out entries, changes made to entries in different handwriting and so on. There are also many 'errors' or categories not 'properly' completed.[4] Read as a text then the manuscript forms reveal how individuals were not simply counted but the categories of the census contested, often resulting in problematic, unexpected and unrecognisable responses. Indeed, during the verbatim transcription of these records a special 'notes' field was created for entering text such as corrections and additional notations. Collectively, the forms reveal an engagement with census taking and their role, along with that of the enumerator, in mediating the interaction and relation between the individual and the state in the making of population.

1.12 A second encounter that informs this inquiry was with textual materials on the taking of censuses, often referred to as contextual data. As part of the same project (<http://www.canada.uottawa.ca/ccri/CCRI/index.htm>) referred to above, I have been involved in the construction of a database consisting of some 12,000 newspaper articles containing narrative accounts about census taking. Like the manuscript forms, the newspaper narratives mediate the relation between the state and the individual. They describe how the wellbeing and governing of the state depends on knowing the population, which requires identifying and assembling each individual into a whole. Census taking is represented as a neutral and official way of knowing and the challenge for the state is to develop and master the techniques necessary to capture the real existing population, both its absolute size (how many?) but also its characteristics (what kinds?):

'But the census has far wider uses than to fix electoral representation. It constitutes, in fact, under the modern system, nothing less than a great periodical stocktaking of the Canadian people, designed to show from the widest angle the point that has been reached in the general progress of the nation. Fundamentally, the importance of the census hinges upon its analysis of the human element or man-power of the country. Their numbers, sex, age, occupation, racial origin, language, education etc. etc. are facts in themselves of the greatest moment. The well being of the state-physical, moral, economic-(including such varied phases as birth and death rates, education, transportation facilities, financial conditions, etc.), with its converse in any form, can be apprehended and interpreted only through the medium of population statistics. Even if the census went no further it would be the basis of all study of our social and economic conditions. Linked with other official data, however, it rounds out the scheme of information by which the Government directs the national affairs. Without the Census, it is literal truth to say that legislation and administration would be carried on in the dark. As the practice of nations in regard to Census-taking tends more and more to uniformity, the Census affords the inestimable benefits of comparison with other countries and enables our national problems to be studied in their general setting.' (Red Deer Advocate, 3 June 1911)

1.13 For the individual, identification is expressed as a responsibility and duty of citizenship, and that for the good of the collective one must be counted and incorporated as a member of the population. Not only is one a part of the whole, but the good of the whole depends on being counted and incorporated as a member:

'It is generally the woman of the house who is interviewed, and we women would be very much insulted if any one accused us of not being patriotic. Here is one way that we can prove that we are. Government is not going to all this trouble and expense for nothing. Their object is to help us, and especially to bring help to the unemployed. Let us assist them' (The Globe, 17 June 1931, p. 17).
'In the opening weeks of June every family and every home in Canada will be visited by a representative of the Government for the great national purpose of taking the seventh census of Canada. The census is really a stocktaking. That is, it provides the information to enable the Government and others interested in the development of the country to formulate progressive policies for the happiness, comfort and prosperity of our people.' (Minister H.H. Stevens, Dept. of Trade and Commerce) (The Sudbury Star, 30 May 1931, p. 7)

1.14 The state's identification of the population is also described as a product of a political process and struggle with consequences for rights, representation and resources. Census taking is shaped by political interference and gerrymandering (e.g. political nepotism in the appointment of enumerators) or is the consequence of political action or inaction (e.g. government policy). Narratives challenge the myth of the census as a neutral document and represent the census as a political tool of the state:

'The manner in which the census is being taken in Calgary is now approaching a public scandal. The census commissioner has practically completed the work, yet it is safe to say, declare citizens, that there are thousands of names which have been left off the lists. In response to appeals in the newspapers, hundreds of persons are coming forward declaring they have never seen a census enumerator…. The Dominion government is being roundly scored, and it is even hinted that the present situation was designed for the purpose of a general scheme to cheat the west out of its proper parliamentary representation.' (Saskatoon Capital, Oct 19, 1911, p.1.)
'In the circle of office holders on Parliament Hill there is no surprise. What the figures would be was pretty well known to anyone having such a thing as a political pull. The Scott government did not want any increase in the provincial representation from the north of Saskatchewan and a true return would have compelled re-apportionment for all the country from Saskatoon north.' (Saskatoon Capital, Oct 19, 1911, p.1.)

1.15 For the individual, the relation is also represented as political in that census taking is a state intrusion into the private lives of individuals whose identification is something to be protected from the prying eyes of the state. At the same time, the individual's identification with census categories and being recognised as part of the population is understood as a form of political recognition. The case of the unrecognised 'racial origin' category of 'Canadian' or the exclusion of housework as an occupation are two such instances:[5]

'People who affect to elect a representative government owe no allegiance to an inquisitorial [oligarchy] who cannot in the very nature of things obtain a mandate from the people to hunt up, classify, specify, tag, label and officially register a pedigree of each person in the land.' (The Voice, 7 July 1911, p. 3)
'The Status of the Homemaker: Resolved that this convention of the W.I. [Women's Institute] requests the Census Department of the Dominion Government to recognize our homemakers as such, when taking the approaching census, and not classify our busiest women as being "no occupation".' (The Globe, 31 December 1931, 15)
'Who and What Are We? Is there such a person as a Canadian?.......Yet when the children of these 'New Canadians' proudly say: 'I am a Canadian', the Government of Canada officially tells them they are mistaken; they are Germans, or Russians, or Poles, or Austrians, or Italians, but they are not Canadians and never can be, nor can their children or children's children ever be throughout the ages yet to come. Pick up the official census returns for all past decades and see if you can find that there is a Canadian living in Canada. Is it not high time that this obsolete, antiquated classification of by-gone years be discarded? Has the time not come when a person born in Canada has a right to be classified as a Canadian and not be compelled to accept official designation of some other nationality which he will never accept for himself personally?' (Banff Crag and Canyon, 2 July 1921, 2).
These are but a selective sampling of the variety of narratives that also mediate the relation between the state and the individual in making population. Like the census manuscript forms, the newspaper narratives represent how both the state and the individual identify through the census, and the two sides of what I call double identification.

'I is; therefore I am': Census Taking as a Practice of Double Identification

2.1 The taking of the census does not simply involve the 'counting of noses', or 'taking stock' and knowing 'how many'. Unlike its precedent - populousness, which was concerned with numbering according to a hierarchical differentiation of different orders of people - making population involves the 'creation of relations of equivalence among heterogeneous entities' to constitute a whole (Curtis 2002: 508).[6] At its most basic the census establishes 'practical equivalences' among subjects whose most general equivalence is to be classified as a member of a population, an undifferentiated abstract essence that effaces individual variation (Curtis 2001). The most general equivalence is referred to as the 'census population base', and defines which bodies count and are to be counted. For example, it often includes only those individuals who are 'usual' residents at an address and excludes visitors.

2.2 The most general equivalence thus involves classifying individuals. But beyond this most basic classification, generalising the individual into the population involves further classifying and identifying her difference and resemblance to numerous categories (male, female, married, single etc.) in relation to a pre-formatted classification grid (sex, marital status, racial origins etc.). The population is thus understood as an entity divided and differentiated into numerous categories and census taking involves identifying each individual in relation to these categories. For the state it is a totalising and objectifying technique such that when all categories are assembled (genders, origins, occupations, incomes etc.) and equivalences translated into numbers the abstract entity called population comes into being or is constructed.

2.3 It is thus through categories or different classes of equivalence that the individual passes from their singularity to a generality (Desrosičres 1998). Identification is not in relation to an abstract essence or statistical object, which is what happens when census officials construct population. Rather, it is through their identification with categories that subjects are identified and identify as members of a population. Categories are 'conventions of equivalence, encoding, and classification, [that] precede statistical objectification' and are the 'bonds that make the whole of things and people hold together' (Desrosičres 1998: 236). It is the categories of the census that connect the individual to the population, hold people together and collectively, once assembled, constitute the population.

2.4 The nominal census manuscript is a paper technology that so constructs population by recording individual names in relation to general 'every person identifiers' (census categories).[7] While there is a long history of different practices of enumerating, the nominal enumeration is a relatively new technique. The first nominal census enumeration in England was in 1841, the United States in 1850, and in Canada 1852 (Curtis 2002).[8] The nominal census is now recognised as an 'essential feature that distinguishes a census from other forms of population accounting' (Goyer and Domschke 1992: 1). Names and addresses are the only individual identifiers on the census manuscript form. All other identifiers are recorded as categories that establish equivalences between bodies such that individual differences disappear and bodies become elements that can be categorised, recategorised, manipulated, and transformed. Census categories thus make individuals comparable and commensurable in all their difference, and can be conceived as part of a general democratic thrust to identify common personhood and equalise subjects (Joyce 2003).

2.5 If we return to the taking of the census, we must take into account this object, that is, the original census manuscript form through which individuals were identified by the state and through which they identified with the population. What we see is that the individual identifiers of name and address are juxtaposed alongside the every person identifiers on the grid that is the manuscript form. It is only when the population is assembled that names are removed and addresses reduced to census administrative districts. Why is this significant? The technique of accounting for every individual's name and address in relation to 'every person identifiers' (categories), I would argue, is bound up with the creation of the 'census subject'. This is a political subjectivity that Joyce argues was necessary for knowing, identifying and then governing population (Joyce 2003). It is similar to Osborne and Rose's (1999) description of the creation of 'opinioned or opinionated people', which was part and parcel of the creation of the technology of public opinion research in the early twentieth century. Drawing on Hacking's study of recovered memory, they argue that genealogies of a research technology can be paralleled with a genealogy of persons: the phenomenon of public opinion was created by the knowledge practice of public opinion research. People 'learn' to have opinions, became opinioned or opinionated and thus opinion polls 'make up' people. In a similar way, through the technology of every person identifiers of the nominal census, people learn to identify in relation to categories or every person identifiers and be census subjects.[9] They learn to think of themselves as part of the population through their identification with census categories. Census categories are thus specific ways of encoding and directing the narratives of the subject who is not a mere informant, but one who has the capacity to generalise, conceive of themselves as both individual and general. It is a capacity inculcated through many such practices of identification of which census taking is just one.

2.6 But then what does identification with a category mean?

I have chosen to frame an answer to this question by thinking about this ostensibly minor comment because it sums up a large volume of textual material and I suggest captures what identification with a census category means. First, the comment acknowledges the subject through a declarative statement - 'I is'. It suggests that the individual is not disciplined or subjected to the census (connoting disciplinary power) but is subjectified by it, that is, made into a census subject. This is the meaning of subjectification that Foucault articulated in The Subject and Power (Foucault 1983). Subjectification recognises subjects as being capable of reflection and self-formation and objects of pastoral power where their subjection is also bound up with struggles against direct domination. Indeed, this is the key link connecting individualising techniques and totalising procedures. The aim of pastoral power is not only to look after the whole community but each individual in particular. The modern state is not above the individual ignoring her existence and who she is but seeks to integrate and shape her, to help her know who she is. This requires exploring the soul, mind and conscience of and producing the truth about the individual both to herself and others. The aims, techniques and methods of subjectification, which emerged with the spread of confessional technologies or more generally technologies of the self, are thus different and must be distinguished from those of objectification, which emerged with the spread of disciplines (Dreyfus and Rabinow 1983).

Figure 5.

2.7 Indeed, if the census only involved making or constructing population then it would be an easier endeavour. The state and its enumerators could simply do the work of identifying, classifying and categorising bodies on manuscript forms and it is likely that in some cases this is exactly what occurs.[10] Enumerators would not experience the frustration and confusion that is often expressed as they try to engage individuals in the categorisation of their existence. The state needs to and wants to find subjects who can identify with its census categories, it needs subjects to tell the truth about themselves, and it needs to affirm that they can recognise themselves as part of the population. This was no more evident in 1911 than in the taking of what was declared to be the first 'scientific' enumeration of 'Indians' and 'Eskimos', the Aboriginals inhabiting the Canadian Far North.[11] Aboriginals could not identify with the categories circulated by the census and see themselves as part of the whole. The difficulties were not simply a matter of language and translation. Cultural concepts for representing divisions of time and social relations also intervened, shaped and confounded enumerators. Thus, Aboriginals could not be classified or identified, nor could corresponding relations be established between them and others in the Canadian population. In other words, census taking was not able to bring forth the subjectivities necessary to produce population in the Far North. Consequently, much effort is expended in educating people, training enumerators, creating instructions, and establishing classifications and categorisations through which individuals can identify.

2.8 Census taking can thus be conceived of as part of the assemblage of subjectifying technologies through which individuals examine and identify who they are in relation to others in the population. To do so individuals must engage in both creative and confessional acts that involve comprehending and identifying themselves in relation to categories of the population. The acquisition of the cognitive tools of generalisation is a necessary precondition of statistical reasoning not only on the part of the state as Desrosičres (1998) well notes, but also on the part of subjects. It is a capacity that involves 'articulation work' - all of the juggling of meaning that goes along with the task of interpreting categories and then performing in the face of uncertainty (Bowker and Star 1999: 310). Rather than an objectifying procedure to control bodies, census categories are specific ways of encoding and directing the articulation work of the subject, and in this way one of the three distinct modalities of the exercise of power: relations of communication (Foucault 1983).[12] Significantly, it is transported to the individual through relations of communication, the paper technology of the census manuscript form. When transported back to the state the forms are interpreted by specialists (the compilers and statisticians of the census authority), the subject's 'interpretive Others', who alone can then reveal the truth about the population and exercise objective capacities. But such truth is only possible through both processes of objectification and subjectification. That is, the subject must be active and able to acknowledge and recognise the truth of the interpretations of specialists (Dreyfus and Rabinow 1983: 180).

2.9 Thus understood, census taking is a liberal form of government in that it requires an active, 'free' subject who is not a mere informant or recipient of census taking but one who has the capacity to shape both its structure and meaning and whose political imaginary includes the population. This capacity is similar to that which Nikolas Rose (Rose 1991) discusses in his account of the rise of the power of numeracy in modern political culture. He argues that the power of numeracy requires subjects who can 'calculate about power', that is, a numerate population that can both comprehend and govern themselves by numbers. Liberal forms of government must keep subjects numerate and calculating, which are prerequisites of the exercise and justification of governing by numbers. Patrick Joyce develops a similar argument when he investigates the 'cognitive possibilities available at the time' that enabled people in the nineteenth-century liberal city to be 'identified as individual and collective objects and subjects for governance' (Joyce 2003: 22).[13] Following this line of thought census taking is part of the cultivation of a political subjectivity that includes the capacity to understand the population as consisting of individuals identified by and organised into distinct categories. It is only when subjects have this capacity that they can associate with and become members of the population. In other words, associating with others by identifying with categories is necessary to make and know the population.

2.10 What then can we say about the categories individuals identify with in their practical everyday lives or in other contexts and those circulated by the census? What is the relationship between the grounded and particular knowledge of identification and statistical knowledge-power? It is a circular process that is captured in Ian Hacking's definition of 'dynamic nominalism'. A kind of person comes into being at the same time as the kind itself is invented; the category and the people in it emerge at the same time (Hacking 1986). What are the specific processes of dynamic nominalism? The question is not whether categories are real but how they are constituted (Desrosičres 1998, Hacking 1986). They are constituted as a result of battles over truth, debates, controversies, etc. - or what I refer to below as classification struggles. Once settled then the phenomenon can be said to exist and can be investigated and acted upon (and I would add, identified with). This is a historically contingent outcome: some categories are 'discovered' and others are not. Hacking says there are many possible descriptions that are true of the world, but the events that establish the truth of one version close off other equally true versions. This contingency does not disqualify the truth status of the versions of the world we arrive at, but does account for why some things become true rather than others, or why some categories become authoritative and others do not.[14]

2.11 Interpreting census categories in this way directs our attention to understanding how particular categories have triumphed over others and have come to be recognised as authoritative. Once they have emerged they survive if it is possible for the state to do things with them (objectification) but only if subjects can also identify with them (subjectification). It is only when the category involves such double identification (state-subject) that it can be said to be 'real'. What then are the microsociological processes that establish particular census categories as 'authoritative' and how are they related to other 'non-authoritative' categories 'out there'? First, classification systems at root are social institutions; they grow out of and are maintained by social institutions (Bowker and Star 1999).[15] There is no great divide between practical and authoritative classifications but there are often fault lines or fractures when lived experience is ordered against a formal, neat set of categories. All classification schemes have multiple sets of fractures arising from such a tension. But the key point is that authoritative classifications are not separate systems from practical ones. Indeed, over time there usually is a convergence between the two: the 'double process by which information artefacts or formal classification systems and social worlds are fitted to each other and come together' (82). In this regard a classification system is partially constitutive of a social world; on the other hand any given social world generates many loosely connected but relatively coherent information resources and tools used to classify. In this way, classification systems and social worlds are involved in processes of mutual constitution or 'co-construction' (Ibid.). Census categories are constituted by and retain an account of what the census authority has done but also the classification work that has been occurring in the social worlds they seek to classify.

2.12 Additionally, because of the dual nature of identification (state-subject) census categories are objects of classification struggles. Rather than benign information merely collected by states for the general good of the polity as suggested by Higgs (Higgs 2004b), census classifications and categories, like others, have moral and ethical consequences and confer advantage or give suffering (Bowker and Star 1999). Categorisation always excludes, names, labels, and produces 'others' and outsiders and thus is an object of political struggle. On this point it is important to draw a distinction between classification struggles and group struggles as argued by Bourdieu (1988). The former consists of symbolic and conceptual struggles over the categorisation of individuals who occupy similar social positions, and who are thus subject to similar conditions and thus tend to perceive themselves as members of a group. But just because individuals are perceived or classified as a group does not mean they will act as a group as this requires the practical and political work of organising and mobilising. Census categories are thus part of symbolic and conceptual struggles over the classification of individuals, which in turn can mobilise and reinforce group struggles. The two struggles can overlap in a number of significant ways: through census categories groups can be 'nominated into existence' (Golderg 1997) which in turn can reinforce their affiliations and identifications (and of course the reverse is also possible); through numerable mediations between individual actors categories can be modified, subverted and changed; and as studies of census taking in a variety of jurisdictions reveal, through recognised census categories groups can claim or be denied social and political rights (Bowker and Star 1999; Higgs 2004a; Kertzer and Arel 2002; Potvin 2005). Finally, once in circulation, census categories can become 'actants' in that they can mobilise subjects to identify with them, a point I explore in detail elsewhere in the case of the category of a Canadian 'racial origin.'[16] In that example, subjects asserted a category not recognised by the state and which was thus not reported in the census. However, when 'Canadian' was finally recognised and added to the list of possible categories the numbers of people identifying as such increased dramatically such that by the end of the twentieth century the category became the fastest growing 'ethnic' origin group in Canada (Boyd and Norris 2001).

2.13 This dynamic is perhaps most visible when we consider the agent's refusal or inability to identify with the categories circulated by the census. While the census imposes a limited repertoire of categories, agents do not necessarily limit their identifications to it. To be sure, agents can and do refuse identification with authoritative categories or claim different ones than those circulated by the state. They also can and do obfuscate and misreport their identification and use identification as a tactical resource. As such we must also consider census taking as not only a subjectifying technology but also a strategy. That is, the subject can and does engage in different strategies of identification (both intentionally and unintentionally). He can absolutely refuse to identify with the census and any of its categories. While legally and ethically compelled to participate (for the good of the collective he has a responsibility to identify with the population) he can refuse and remain 'outside' of the population. Alternatively, he can identify and be part of the population through his association or solidarity with its categories. To resist or challenge particular categorisations would then represent a classification struggle. The census form is thus a subjectifying technology that is also used strategically and as such always results in categories that are non-authoritative. For while professing completeness and attempting to incorporate all identifications, censuses always produce 'others'.[17] Indeed, the ubiquitous 'other' category attests to the classification struggles over authoritative categories between the state and individuals.

2.14 Returning to the comment quoted above, a second aspect concerning the relation of the individual to census taking emerges. The comment suggests that the census does not capture an essential identity but rather identification in relation to official categorisations of the population made authoritative by the state. It is understood that there are other identifications and 'existences' and that the census is an 'official' one that binds people together into a population. It is in this regard that I consider identification with census categories as a particular type of association and governing practice of the state. Put another way, we could say that identifying with categories of the census is how one officially associates with the population. It is not an association with an abstract, statistical concept but with others, and is political because the association is made with and against others: with others who identify with the same categories and against others who identify with different ones. This is a particular way of thinking that census taking mobilises and reinforces, an understanding that the population is a politically differentiated entity.

2.15 Furthermore, while reducing individual variability to a given repertoire of categories and eventually producing the abstract entity called population, the identification with categories also reveals individuality. While individuality could be expressed by asserting oneself in the category of 'other' as discussed above, this is often an identification still shared with others (as noted above in the discussion of the relation between practical and authoritative classifications). Rather, individuality is revealed in the census because identification is never with one but rather the census demands that the individual simultaneously identify with the categories of numerous classifications: age, gender, marital status, ethnicity, origins, language and so on. As the numbers and diversity of individuals increases and more classifications and categories are asserted and recognised then the more likely a person's identification when strung together on a census manuscript form will constitute a particular individuality. Even when the string of data is stripped of a person's name and specific address (unique identifiers) the particular combination of every person identifiers is often unique, especially for individuals who have highly visible or 'disclosive' identifications (e.g. such as minority ethnic and occupation categories, or residence in small, remote geographical areas).[18]

2.16 This echoes what Simmel observes when he writes, 'the larger the number of groups to which an individual belongs, the more improbable is it that other persons will exhibit the same combination of group-affiliations, that these particular groups will 'intersect' once again in another individual' (Simmel 1955: 140). Individuality through the census can thus be considered as an identification with a unique combination of census categories. However there is a tension in this constitution of individuality. An identification is simultaneously individual and collective as it comes into being through the identification with generalised categories that are shared amongst a large number of people. It is a subjectivity that thus has affinities to Simmel's definition of the stranger: 'The stranger is close to us, insofar as we feel between him and ourselves common features of a national, social, occupational, or generally human, nature. He is far from us, insofar as these common features extend beyond him or us, and connect us only because they connect a great many people' (Simmel 1950: 421). Through census categories we thus articulate our common features, a general similarity, through which we can identify and associate with others and thus be 'close' to them. At the same time, our common features are shared by a large number of other people and thus make us 'distant' from them for 'what is common to two is never common to them alone' (Simmel 1950: 422). Categories are common to many and thus our census identification is a particular type of stranger to whom we are both near and far.

2.17 In this regard census taking captures the tension between individuality (subjectivity) and generality (objectivity) that Simmel so well described as key to modern culture. Like the crowd the census can be conceived as a form of association that brings us close to others and at the same time makes us distant. And just as the stranger who passes us in the crowd generalises about who we are, so too does the manuscript form. When completed, the form visually represents us in our generality and individuality and perhaps uniquely makes it possible to hold both conceptions simultaneously. But, its completion involves numerous agents who mediate the relation between the state and the individual - the manuscript form, the enumerator and newspaper media. It is through their mediation that census taking produces population, that is, contributes to the development of the capacity to recognise oneself as a member and part of a population.

The Ontological Politics of the Census

3.1 Social scientists fret over the meaning of census data and numbers. What are they evidence of? State power? Processes of objectification? Constructed equivalences? None of these accounts helps us much when it comes to 'using' the data in research projects or government programs. Simply acknowledging that data is the product of messy power relations, influences, and processes of construction does not go a long way. What would an alternative understanding look like?

3.2 I have argued that the census produces more than statistics, numbers and the abstract entity of population. It is through processes of double identification (state-citizen) that census categories come into existence, become facts and can then in turn not only be measured, analysed and assembled (objectification) but also be identified with (subjectification). The presence of such double identification makes an ostensible division between facticity and representation artificial. While the foregoing has focused on the 'census moment' - that is, when the census is taken - there are indeed many other moments when the census is active in this production. Census statistics are deployed in numerous government policies, mobilised as evidence of the state of the population in political and media debates, appealed to in group and individual claims to social and political rights, and used by individuals to assess their relative status and life situation. In all these ways census statistics both produce population and come to reinforce the 'reality' of census categories. As Law and Urry (2004) argue in relation to the social sciences, in these ways methods do not simply describe the world as it is, but also enact it. Methods are performative, have effects, make differences, enact realities, and can help to bring into being what they also discover. The challenge when analysing the census then is to reveal its performativity rather than accept the state's version of what the census enacts. Instead, an ontological politics of the census demands that we question the social realities and worlds that the practice legitimises.

Author Note

A draft of this paper was presented to a workshop of the Centre for Research on Socio-Cultural Change (CRESC) < http://www.cresc.ac.uk/ >: Narrative, Numbers and Social Change at the University of Manchester, 1-2 November 2007.

Notes

1 At the time of the introduction of censuses in mid nineteenth century Canada, no other state practices subjectified individuals and constructed population in this way. No system of civil registration existed and the collection of information on vital events was the purview of local clerks, who unevenly and variably compiled information from religious officials and coroners (Curtis 2001: 68-69). However, by the late nineteenth century most provincial jurisdictions had introduced a system of civil registration and together with censuses constituted the main practice through which the state identified, categorised and constructed the population.

2 The Canadian Century Research Infrastructure Project (CCRI) is a five-year interdisciplinary and multi-institutional initiative involved in building a set of interrelated databases concerning the 1911-1951 Canadian censuses of population. I am one of eleven Team Leaders from seven universities across Canada and co-direct the York University Centre with Professor Gordon Darroch. The project involves the construction of databases including digitised data from original census manuscript forms, and documentary data derived from Statistics Canada files, newspaper commentaries, and parliamentary and legislative debates. For further information see a recently published special issue of Historical Methods (2007): 40(2).

3 Geocoding involves assigning unique geographical identifiers to each sampled individual thus enabling spatial analyses and mapping of microdata. In the case of the CCRI, this involved creating polygons for nearly 32,000 census subdivisions (the basic spatial unit for census administration in the 1911-51 census years). Each sampled individual in the microdata is then linked to a specific polygon in a geodatabase. See discussion in (St-Hilaire et al. 2007)

4 Thousands of such pages were so produced in the course of census taking, but due to confidentiality laws, these have been stored away on microfilm in government archives only to be seen by the occasional researcher. Scholars interested in censuses have long fretted over the problems that these original census manuscripts present and have identified coding and translation rules for standardising the idiosyncrasies, unpredictability, uncertainty and variability. However, they have not engaged with the uncertainties apparent in the manuscripts as an invitation to rethink both what is being recorded and represented, and the agencies and capacities that census taking developed and mobilised.

5 Until 1986, the census discouraged and advised against 'Canadian' as a category in the classification of racial or ethnic origin. However, numerous respondents did indeed report 'Canadian' as their origin and by 1971 Statistics Canada for the first time reported their numbers: over 71,000 respondents insisted on Canadian as their single response (Boyd 1999). For further discussion see (Ruppert 2007).

6 Curtis outlines the long history of the concept of populousness, which he argues was based on a logic that hierarchically differentiated between orders of the people in relation to their 'essences': knights fight, priests pray, peasants till.

7 See Caplan and Torpey (2001) for a discussion of how 'every person identifiers' enabled linking observable regularities to individuals.

8 The unit of enumeration in earlier censuses was the household. In England the nominal return was introduced in 1841 as part of the state's centralisation of administrative arrangements and as a way to monitor and govern the conduct of enumerators. The colonies were urged to follow suit and conduct enumeration by name to ensure accuracy (Curtis 2001: 92-93).

9 Newspaper editorials in the early twentieth century often lamented and chastised those individuals (especially 'foreigners') who did not comprehend how to articulate themselves properly into census categories. Knowing how to identify with categories of the census was often expressed as a citizen's obligation.

10 Some manuscript forms reveal a consistency in the completion of categories that suggests this is sometimes the case.

11 The examples are more fully discussed in Ruppert (forthcoming).

12 Foucault conceived of subjectification as involving the exercise of relations of power (that existing between individuals or groups), relations of communication (that involving the transmission of meaning) and objective capacities (that exerted over things) (Foucault 1983).

13 He specifically identifies the invention of the Penny Post in 1840s Britain, which linked names to addresses and residences and accelerated the numbering of house doors and fixing of street signs. The Penny Post, which replaced delivery that was dependent on local knowledge and face-to-face contact, can be understood simultaneously as an individualising technique (whereby each person was attached to a particular house address) and as a totalising technique (whereby all houses and names were circulated and assembled into the city). People and their houses thus could be identified as individual and collective subjects and objects of governance.

14 Curtis (as do other researchers interested in this question) argues that census making must also 'reflect social relations' (Curtis 2001: 34). He notes, for instance, that 'statistical knowledges are conditioned by the materiality of the social relations they attempt to appropriate; they are historically specific knowledges that are adequate to particular kinds of social objects and, by implication, inadequate to others' (308). The latter, Curtis says, feeds off other ways of configuring and knowing social relations and in turn come to shape those very social relations.

15 Bowker and Star (1999) address this question in their discussion of the relationship between formal or scientific classification systems, and informal or practical classification systems, deployed in the everyday. They cite in particular Mary Douglas's work on how practical classifications of the everyday become reified, and Durkheim and Mauss, for whom primitive, practical social classifications are linked to the first scientific classifications.

16 See (Ruppert 2007). See also note 5.

17 See (Ruppert 2007), (Ruppert forthcoming) and ( Bowker and Star 1999) for discussion of how censuses always create 'others' and marginal people. As far as I know, a genealogy of the 'other' category is yet to be written. In the 1951 Canadian census, however, for the first time categories were listed on the form and an 'other' box was included for enumerators to write-in categories not listed. Previously, as noted above, the form was blank and enumerator instructions provided direction on recognised categories. Responses that did not fit in categories (and usually too few in number) were then recorded as 'other'.

18 This is especially revealed when researchers deal with confidential census microdata (individual level data), which demands not only removing names and addresses, but also applying a number of statistical disclosure control measures to reduce data specificity such as distorting the data, decreasing sampling sizes and rounding. See (Purdham and Elliot 2007) for summary and discussion of disclosure risk in samples of anonymised records.

References

BOURDIEU, P. (1988) 'Social Space and Symbolic Power', Sociological Theory, Vol. 7, No. 1, pp. 14-25.

BOWKER, G. C., and S. L. STAR. (1999) Sorting Things Out: Classification and Its Consequences. Cambridge, Massachusetts: The MIT Press.

BOYD, M. (1999) 'Canadian, Eh? Ethnic Origin Shifts in the Canadian Census', Canadian Ethnic Studies Journal, Vol. 31, No. 3, pp. 1-19.

BOYD, M. and D. Norris (2001). "Who are the "Canadians"? Changing census responses, 1986-1996." Canadian Ethnic Studies Journal 33(1): 1-26.

CAPLAN, J., and J. TORPEY. (2001) 'Introduction', in J. Caplan and J. Torpey (editors) Documenting Individual Identity: The Development of State Practices in the Modern World. Princeton, NJ: Princeton University Press.

CURTIS, B. (2001) The Politics of Population: State Formation, Statistics, and the Census of Canada, 1840-1875. Toronto: University of Toronto Press.

CURTIS, B. (2002) 'Foucault on Governmentality and Population: The Impossible Discovery', Canadian Journal of Sociology, Vol. 27, No. 4, pp. 505-533.

DESROSIČRES, A. (1998) The Politics of Large Numbers: A History of Statistical Reasoning. Translated by C. Naish. Cambridge, Massachusetts/London, England: Harvard University Press.

DREYFUS, H. L., and P. RABINOW. (1983) 'The Genealogy of the Modern Individual as Subject', in H. L. Dreyfus and P. Rabinow (editors) Michel Foucault: Beyond Structuralism and Hermeneutics. Chicago: University of Chicago Press.

FOUCAULT, M. (1983) 'The Subject and Power', in H. L. Dreyfus and P. Rabinow (editors) Michel Foucault: Beyond Structuralism and Hermeneutics. Chicago: University of Chicago Press.

FOUCAULT, M. (1994). The Political Technology of Individuals. Power: Essential Works of Foucault 1954-1984. J. Faubion. New York, The New Press. 3: 403-417.

GOLDERG, D. T. (1997) Racial Subjects: Writing on Race in America. New York: Routledge.

GOYER, D. S., and E. DOMSCHKE. (1992) The Handbook of National Population Censuses: Europe. New York: Greenwood Press.

HACKING, I. (1986) 'Making up People', in T. C. Heller, M. Sosna and D. E. Wellbery (editors) Reconstructing Individualism: Autonomy, Individuality, and the Self in Western Thought. Stanford, CA: Stanford University Press.

HIGGS, E. (2004a) Life, Death and Statistics: Civil Registration, Censuses and the Work of the General Register Office, 1836-1952. Hatfield, Hertfordshire: Local Population Studies.

HIGGS, E (2004b) The Information State in England: The Central Collection of Information on Citizens since 1500. Hampshire, England: Palgrave MacMillan.

IACOVETTA, F. AND MITCHINSON, W. (1998) 'Introduction: Social History and Case Files Research' in On the Case: Explorations in Social History, (eds) F. Iacovetta and W. Mitchinson, University of Toronto Press: Toronto.

JOYCE, P. (2003) Rule of Freedom: Liberalism and the Modern City. London: Verso.

KERTZER, D. I., and D. AREL. (2002) 'Censuses, Identity Formation, and the Struggle for Political Power', in D. I. Kertzer and D. Arel (editors) Census and Identity: The Politics of Race, Ethnicity, and Language in National Censuses. Cambridge/New York: Cambridge University Press.

LAW, J., and J. URRY. (2004) 'Enacting the Social', Economy and Society, Vol. 33, No. 3, pp. 390-410.

OSBORNE, T., and N. ROSE. (1999) 'Do the Social Sciences Create Phenomena?: The Example of Public Opinion Research', British Journal of Sociology, Vol. 50, No. 3, pp. 367-396.

POTVIN, M. (2005) 'The Role of Statistics on Ethnic Origin and 'Race' in Canadian Antidiscrimination Policy', International Social Science Journal, Vol. 57, No. 183, pp. 27-42.

PURDHAM, K., and M. ELLIOT (2007) 'A Case Study of the Impact of Statistical Disclosure Control on Data Quality in the Individual UK Samples of Anonymised Records', Environment and Planning A, Vol. 39, pp. 1101-1118.

ROSE, N. (1991) 'Governing by Numbers: Figuring out Democracy', Accounting, Organizations and Society, Vol. 16, No. 7, pp. 673-692.

RUPPERT, E. (Forthcoming) 'Becoming Peoples: 'Counting Heads in Northern Wilds'', Journal of Cultural Economy.

RUPPERT, E. (2007) 'Producing Population.' CRESC Working Paper Series, Paper No. 37. Milton Keynes: The Open University. < http://www.cresc.ac.uk/publications/papers.html >.

SCOTT, J. C. (1998) Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed. New Haven and London: Yale University Press.

SIMMEL, G. (1950) 'The Stranger', in K. Wolff (editors) The Sociology of Georg Simmel. New York: Free Press.

SIMMEL, G. (1955) 'The Web of Group-Affiliations' (editors) Conflict and the Web of Group-Affiliations. New York: Free Press. Original edition, 1922.

ST-HILAIRE, M., B. MOLDOFSKY, L. RICHARD, AND M. BEAUDRY (2007) Geocoding and Mapping Historical Census Data: The Geographical Component of the Canadian Century Research Infrastructure. Historical Methods 40 (2): 76-91.