1 – A scientific and political issue

11When Louis Henry joined INED in October 1946, just one year after the founding of the Institute, his career had been far removed from historical preoccupations. Born in 1911, educated in military schools, Henry had been admitted to the École Polytechnique in 1931. Like many students of the 1930s classes (Guigueno, 1994), he graduated with officer rank. Commissioned in the artillery, he was a prisoner for the duration of the war. After a brief stint back in the army, he was hired by INED while having no expertise in demography — let alone history — through the good offices of his fellow alumni of Polytechnique, Jean Bourgeois [3] and, most of all, Paul Vincent. Although they were his contemporaries, Bourgeois and Vincent had already some experience in demographic research, acquired during the Occupation in the small unit dedicated to the discipline at the Fondation Carrel [4].

12When Vincent, and later Bourgeois began their research, the outlook in the profession was predicated — as it had been between the wars — on the dreaded downward trend in the birth rates of western nations. Although Henry acquired the same theoretical references as his two ex-classmates [5], his thinking was rooted in a totally different context: that of the mysterious upturn in the birth rate, as revealed by none other than Bourgeois himself. At the outset, everyone regarded it as a transient phenomenon, comparable to the one that characterized the two years that followed World War I. But after the high birth figures of 1946 and 1947, which received wide press coverage, the trend persisted into 1948. Henry voiced his surprise in a series of lectures. With his friend Jean Bourgeois, but against the opinion of Alfred Sauvy [6], he came to the view that “fertility has risen very sharply from its pre-war level” and set out to find the explanation.

13In the following years, this euphoria often gave way to uncertainty and hesitation. The year 1949 was again greeted with apprehension. Demographers feared that the 1919-20 scenario might yet repeat itself. The catch-up might simply have lasted longer this time, because couples had been separated longer. Again, the facts “disprove this more or less openly articulated expectation […]. Everyone, of whatever opinion, would like to know the causes of this influx of newborn and to know whether it is likely to last” [7]. Had the exhausted western world overcome its reluctance to procreate and chosen life again? The question of whether fertility had actually increased became a subject of polemics for years. The difficulty of providing a clear-cut answer is not hard to imagine, given the exceptional conditions that had characterized the war and its aftermath.

14Louis Henry was directly exposed to this unprecedented demographic suspense. In June 1948, he was assigned to the computation of “family projections” — a subject he was never to abandon and that would lead him to population history. At the same time, Alfred Sauvy gave him responsibility, alternating with his friend Bourgeois, for “La situation démographique”, a quarterly summary description of the French demographic situation published in the journal Population [8]. Henry thus monitored — practically as it was occurring — the unpredictable course of the birth curve, in particular by means of a convenient indicator, the number of “pregnancy cards” issued in Paris.

15Louis Henry’s intense preoccupation with the issue was due to the modus operandi of INED, whose remit was to provide hard demographic intelligence, especially to the government (Rosental, 2003). Sauvy and his influential Technical Committee (the equivalent of today’s Scientific Council) explicitly gave precedence to “applied” over “basic” research [9]. Thus for Henry the theoretical issue was simultaneously a public policy issue. The reality of the baby boom would determine the appropriateness of massive investment in public or private infrastructure, most notably schools and housing. Admittedly, the link was less mechanical than might be assumed. The evidence of a sustained increase in fertility justified capital investments, which Louis Henry advocated. But at the same time, Henry shared the fear of a “Malthusian reaction” — be it the alarm in English-speaking countries about population growth in the developing world, or the opposition in French “conservative circles” to the expenditures that population growth would induce. These fears were expressed repeatedly, most notably by INED’s director, Alfred Sauvy, but also by his research staff [10]. In this case, professional expertise was inseparable from pronatalism, a position publicly advocated in the Institute’s name [11]. Henry supported and participated in this “anti-Malthusian” struggle. Through lectures, newspaper articles, rejoinders to journalists, and memoranda to senior civil servants (Fonds Henry, art. 8), he sought to minimize the link between the demographic surge and the spending on infrastructure.

16In regard to family allowances, the issue took a paradoxical turn. The system — developed in the 1930s and later under Vichy, and greatly extended after the Liberation — was the standard bearer of French “voluntaristic” demographic policy. But INED researchers feared that it contained the seed of its own destruction. Designed for a context of depressed birth rate, it would become a source of runaway spending if the birth rate recovered. This was a cruel dilemma: a system threatened by its own efficiency. The uncertainty about the population trend was thus compounded by uncertainty about its financial implications and, most of all, its political fallout.

2 – From forecasting to history

17In this context, forecasting took on a strategic dimension. Both the “erratic fluctuations of the gross and net reproduction rates in the past ten years” (Clark, 1949) and the advent of new analytical tools fragmented the perception of the demographic situation and caused demographers to lose their bearings. The immediate postwar years represented a complete reversal in comparison with the early days of demographic forecasting. Since the 1920s, population experts tended to emphasize the predictability of phenomena and would extrapolate current trends into the future (Hajnal, 1947b). In France, pioneer Alfred Sauvy actually feared the perverse effect of the momentum of population phenomena: why would political rulers finance population policies whose effects would be felt only in the very long run (Sauvy, 1928, 1929 and 1932)?

18The uncertainty surrounding the unexpected baby boom shook this confidence. In the United States and in Britain, well-funded studies during the war were premised on the unquestioned assumption that the depressive trend of the 1930s would persist (Notestein, 1944). Barely had their first findings become known that the evidence of an upswing in the birth rate became undeniable. This shock (Henry, 1966) was accompanied by a loss of innocence, with many demographers now recognizing that research sponsors were more interested in obtaining scientific legitimacy for their projects than guidance for enlightened policy-making (Cox, 1954). The harshest self-criticism came from the British demographer John Hajnal. After attacking the concept of demographic forecasting at the 1954 World Population Conference in Rome, he questioned the very usefulness of the profession. How could its existence be justified if the sophisticated techniques for the analytical decomposition of demographic phenomena performed no better, in the area of forecasting, than the rudimentary tools of the interwar years (Benjamin et al., 1955; Henry, 1966)? Demographers’ predictions, Hajnal argued, had no scientific value. At most, by providing a basis for population policies, they could have self-fulfilling effects.

19In the mid-1950s, Louis Henry corresponded with Hajnal. In a letter of 29 December 1955, he explicitly defended demographic science as an indispensable guide for action. Yet he too had drawn lessons from the surprise of the baby boom and the instability of the previous two decades. Henry was simply more “constructive” than his interlocutor in his cautious acceptance of forecasts that extrapolated current trends to the very short term [12]. He based them not on theory but on “empiricism”, in one of the very rare writings in which he refers explicitly to his experience as artillery officer.

20Hajnal countered with a totally sceptical view [13]. Yet, beyond their disagreement, the main point was that for both men, the fallibility of forecasting enhanced the value of understanding past phenomena, the main difficulty being in demonstrating their importance to paymasters misled by an erroneous concept of “effectiveness” [14]. Hajnal (1955) stated publicly that one of the few good ways in which one could claim to talk about the future was by studying the past. Likewise, the reference to the past formed the leitmotiv of Louis Henry’s many public lectures to non-specialists on the French demographic situation [15]. This response to an audience eager for forecasts was the first source of Henry’s interest in the demography of times past.

3 – Introducing nuptiality

21In the interwar years, most authors discussed each country’s fertility as a whole. Demographers had concentrated their attention on overcoming the bias linked to the age distribution or, as Alfred Lotka, on integrating fertility in a general dynamic theory of population. Few had thought to distinguish “marital” fertility from “illegitimate” fertility, and in any case, only a tiny minority of births in western countries occurred out of wedlock. The methods they had proposed took on an entirely new relevance at the end of World War II [16].

22In a widely discussed article, Hajnal (1947b) showed that in several countries the upturn in birth rates coincided with an increase in the number of marriages. This trend mechanically increased the proportion of young couples in the overall married population. Now, in these populations practising contraception, births tended to occur in the early years of marriage. Hajnal concluded that demographers were in danger of confusing two different interpretations of the fertility recovery. The first concerned the behaviour of spouses. Couples, it was argued, now wanted a greater total number of children. Owing to the inertia of preferences, the upswing might persist. A second explanation, in contrast, minimized the impact of the phenomenon. In this view, earlier marriage had merely led to a change in the timing of births and caused a temporary rise in the period indices [17].

23Nuptiality thus became an essential element of the postwar demographic scene. This innovation was at the same time regarded as a means of taking into account the male population, whose exclusion from the analysis continued to fuel controversy among demographers [18]. Similarly, the introduction of a cohort approach conferred a new analytical status on the question of contraception. In the U.S., the links between differential social fertility and the use of contraception had been measured empirically in the 1930s, through a series of surveys funded by birth control advocates (Kiser, 1971). After the war, the normative ideology of birth control and the investigation of actual practices converged. Demographers emphasized the importance of couples’ planning their number of offspring (Hajnal, 1947b). The semantic shift in the 1940s from the stern “birth control” to the more euphemistic “family planning” illustrates this new perception, in a broader setting than the scientific community alone (Hartmann, 1997).

24The distinction between populations that were assumed to widely practice contraception, and the other populations, emerged from this phase in a new form. At the time, a strict analytical division separated these two types of population: those in which couples decided on the size of their family in advance, and those in which fertility was not deliberately controlled [19]. We will encounter this dichotomy throughout this article. In the post-war years, demographers believed that contracepting populations had clear objectives, and that timing was the only uncertain factor (Hajnal, 1947b). The speed with which couples reached their fertility targets depended, in particular, on economic conditions [20]. The fertility of populations assumed not to be practising contraception was governed by laws of its own. In a period when concern for regulating population growth in the poor regions of the planet was taking shape, demographic theory legitimated a division between the study of fertility in the industrialized countries and in the developing countries [21].

4 – Family micro-history and demographic macro-changes

25The emphasis on marital fertility and the opposition between contracepting and noncontracepting populations structured the model of historical demography developed by Louis Henry a few years later. But its very purpose — the microscopic analysis of the reproductive behaviour of couples — also originated in the theoretical debates about the baby boom.

26In his 1947 article, one of Hajnal’s targets was the interpretation of the spectacular recovery of German fertility during the early years of Nazism. The political importance of the subject revealed the limits of the “macroscopic” indicators used by the demographers of the interwar years. Adopting a rationalist stance, Hajnal regarded aggregate demographic data and the fertility or reproduction rates that could be derived from them as too imprecise. Their use left too great a margin for personal interpretation and hence for ideology. The British demographer preferred a “microscopic” approach that focused directly on the behaviour of couples. Did German fertility change in 1933 and, if so, was this outcome specific to Nazism? An answer to this question would require determining whether, for an equivalent duration of marriage, German couples of the mid-1930s had more children than those of earlier cohorts or of other countries. The lack of sufficiently detailed data caused the sceptic Hajnal to suspend judgment.

27His reasoning exemplifies the issues and methods that spread after the war (Clark, 1949). Was the upswing registered by fertility indices the result of statistical bias or of a revolution in personal behaviour? The growing use of opinion polls and the introduction of sampling techniques in survey procedures made such investigations possible (Hyrenius, 1948; Shryock, 1950). But because of their cost they were not within everyone’s reach, and many demographers around the world had to continue to rely on national statistical sources.

1 – The impossibility of measuring natural fertility

35For Henry, the “pure” components of fertility are first and foremost physiological, such as the female propensity to conceive during a menstrual cycle. But other components are social or cultural, starting with breast-feeding practices. Subject to extreme individual variations, they determine the pace of the mothers’ return to a new fecund state. This set of factors goes by the name of natural fertility [31]. The expression is misleading, as the notion is both related to biology and conditioned by culture [32].

36Among the factors that Henry labelled as “ physiological”, natural fertility includes the proportion of sterile couples (which varies with age), the length of the infecund period in fertile couples and, most important, “fecundability”. Defined as the probability of conceiving during one menstrual cycle (Henry, 1951 and 1952), this concept was developed by the influential statistician and economist of Fascist Italy, Corrado Gini, with whom Henry corresponded after the war [33]. Henry also cited—not uncritically — the eminent American biologist Raymond Pearl (Henry, 1953a) [34]. By his references to these two authors, complemented by his study of physiology textbooks [35], Henry positioned himself, in the postwar years, in the line of the demographic tradition that is closest to the natural sciences. Indeed, his biologism was sometimes attacked, as Gini’s had been before him (Grauman, 1959). Although not a fervent eugenist, Henry did not hesitate to borrow from this current of scientific thought — which had flourished between the wars — the notions and tools that seemed useful to his research, ignoring or masking their ideological implications. His “pragmatism” toward Gini, whose scientific positions were challenged in the postwar world, is in this respect revealing of a more general attitude [36].

37At the same time, other determinants of natural fertility were sociocultural or, if one prefers, anthropological, such as breast-feeding practices or “customs and taboos” influencing the resumption of intercourse after delivery. If we wanted to subject Henry’s ideas to retrospective and somewhat anachronistic criticism, we would have to emphasize his ultraculturalist view of behaviour at least as much as his biologism [37].

38The notion of natural fertility applies, in the first place, to “non-Malthusian” populations — the term still used in the immediate postwar years to describe non-contracepting populations. But it also applies to “Malthusian” populations, within the confines of a fairly rigid vision typical of the period. Since the couple decides at the outset how many children it wants, it will start with a natural fertility phase, and will then cease to reproduce once it has reached its desired family size. In this second phase, given the less than total effectiveness of its contraception, the couple may conceive unwanted children. At the aggregate level, their distribution in time will bear some resemblance to that of couples who follow the “natural fertility” model. Lastly, the reference to natural fertility allows an estimation of phenomena such as the proportion of prenuptial conceptions.

39By overcoming the great post-war analytical divide between contracepting and non-contracepting populations [38], the concept of natural fertility was suitable, in Henry’s view, as a universal benchmark. From his earliest work — on colonial populations — Henry made empirical use of the notion of a space-time continuum. Always using the most robust statistical series, he looked in other temporal or spatial contexts for populations with characteristics comparable to those of the populations he was studying, on which more data existed. For example, by identifying for Spain in 1880 and for Italy in 1900 two male-mortality curves that enclosed the curve for Algeria in 1936 between them through the adult ages, he used them to “adjust” Algerian mortality at older ages (Henry, 1947). In effect, Henry treated history as a stock of observations comparable with the colonial and developing-region populations [39]. This approach, whose foundations were not made explicit, suffered from the limited number of statistical studies on historical populations that could be regarded as reliable [40]

40It is important to realize the powerful constraint represented by this lack of evidence. At the time, demographers and historians distinguished between “statistical” and “pre-statistical” populations, depending on census availability (Rosental, 1997). As regards fertility, the division was overdetermined by the difference between contracepting and non-contracepting populations. From the nineteenth century, population specialists were aware of the difficulty of studying the mass diffusion of contraception, which partly preceded the era of censuses. But this obstacle took on a new significance in Henry’s generation. U.S. and British studies carried out in the 1930s had shown the effect of contraception on differential fertility by social class. The enigma of the baby boom underscored the importance of the intimate decisions of couples. For the postwar demographers, the opposition between contracepting and non-contracepting populations corresponded to a deep anthropological division, a “break between civilizations” (Chevalier, 1946).

41For pronatalist demographers, the relevance of the question was intensified by nostalgia for lost greatness. France’s influence, they argued, was waning because of the early decline in its fertility at the end of the eighteenth century, several decades before the rest of Europe. At a more personal level, the correspondence between Louis Henry and Jean Bourgeois-Pichat shows the two men’s sensitivity to the worried debates on contraception that were stirring the Church and Catholic circles at the time. Now, faced with the crucial question of dating the advent of contraception, demographers had difficulty extending knowledge of past populations back beyond the Napoleonic period (Bourgeois-Pichat, 1951; Vincent, 1947), and the historian Louis Chevalier had abandoned all hope for a firmly-based understanding of periods prior to the eighteenth century (Chevalier, 1946). Overcoming this time barrier was one of Louis Henry’s greatest achievements.

2 – A Northwest Passage in the archives

42The solution to the problem took shape in 1953. A few months earlier, Louis Henry’s attention had been drawn to a widely quoted article by Pierre Goubert (1952) that proposed a statistical analysis of the seventeenth-century parish registers of the Beauvaisis, a region north of Paris [41]. These systematic registrations of population movements (baptisms, marriages, burials) had been the object of earlier but sporadic scrutiny by historians and local scholars. They had also been the focus of substantial critical evaluation by archivists, included in the curriculum of the École des Chartes, the French school for archivists and paleographers [42].

43It was through his critique of Goubert’s article that Louis Henry first penetrated into the historians’ territory (Henry, 1953c). He designed a method of study that he would try out on three projects. To offset the lack of statistics for older periods, Henry initially relied on documents such as livres de raison (registers kept by heads of households) or genealogies. The opportunity to test these sources was given to him by Michel Fleury, deputy archivist of the département of the Seine, who told him about the data then being assembled on the patrician families of early modern Geneva [43]. Meanwhile, Henry asked his assistant, Louis d’Adhemar, to assess the value of genealogical materials for historical demography. These have the advantage of being, so to speak, “pre-processed”: they provide instant information on the demographic event history of couples and lineages. On the downside, Henry was aware that this information cannot always be corroborated and is socially biased [44].

44The arrival at INED of a young Canadian trainee, Jacques Henripin [45], proved decisive. Henripin was analysing the genealogical data compiled in the previous century by Abbé Tanguay on the French Canadians in the seventeenth and eighteenth centuries (De Tanguay, 18711890; Henripin, 1954). The data were drawn from the old parish registers of Québec, so in helping Henripin, Louis Henry was able to judge for himself the richness of this source and the feasibility of exploiting it. Henripin was able to relate mother’s age at marriage and at the birth of her children, to study birth intervals, to analyse their sensitivity to infant mortality, and even to estimate “fecundability”. Using Henripin’s data, Henry refuted the hypothesis that birth spacing in non-contracepting populations was constant until sterility simply put an abrupt end to family formation. On the contrary, Henry noted that spacing remained constant only until the antepenultimate birth. It then increased slightly, and the interval between the last two births was much wider. The shortcomings of contemporary statistical publications were forgotten. For the first time, Henry had identified a characteristic of “natural fertility”. He was to find it again in very different contexts, and this strengthened his confidence in the concept’s relevance (Henry, 1954a).

45Louis Henry’s interest in parish registers was reinforced by his encounter with Étienne Gautier. This “enthusiastic amateur, by occupation a civil servant in the Trade Ministry, had spent many holidays patiently reconstituting the families of his native village” [46] of Crulai in Normandy. Gautier brought Henry the results of his compilation of data from the pre-Revolutionary parish registers. Their availability and universality made the registers a precious source for investigating the demographic conditions in pre-industrial France. But they raised two series of problems.

46The first was the problem of analysing the data source, and Henry began to tackle this in his 1953 article. He was helped by young historians whom INED hired for him on temporary contracts, and by Michel Fleury (1923-2002). It was with Fleury that Henry published in 1956 the manual that was to give historical demography its true definition, the monograph on Crulai being its first full-scale application (Fleury and Henry, 1956; Gautier and Henry, 1958). Fleury also played a decisive role by putting Louis Henry in contact with the network of archivists interested in parish registers.

47A second, more serious obstacle was the time and energy required to extract data from the parish registers. For each geographic entity examined — in practice, a single parish, given the bulk of the data set — all the baptisms, marriages, and burials for the period of study had to be recorded on index cards; then for each married couple, a “family reconstitution form” (fiche de famille) had to be started [47]. The implementation of the method was predicated on a division of labour in which the designer, Henry, would check and process the data gathered by countless collectors. Their training posed a problem [48], as did the project organization, even though that was probably one of Louis Henry’s strengths [49]. But the greatest difficulty was to find a large enough work force. Three main sources were used. First, local amateur historians who had compiled “unofficial” population histories volunteered their services [50], as did secondary schoolteachers and members of the clergy teaching in Catholic schools. Second, one of the strengths of the Henry method was its accessibility to non-specialists: Louis Henry relied on non-historian volunteers identified by département archivists [51]; Alfred Sauvy and Michel Fleury did their best to get the heads of teacher-training schools (écoles normales) to channel their students toward the study of parish registers as a topic for their final thesis. Third, Louis Henry actively sought help from the community of historians. His initiative was relayed by the Sub-Commission for Historical Demography, which reported to the French section of the International Committee for Historical Sciences [52]. In 1954, Louis Henry publicized his method among historians through a series of lectures. Academic historians sympathetic to his call steered some of their students toward topics in historical demography [53].

48Teachers often delegated full responsibility for supervising these theses to Henry and his team [54]. Not only did the demographer, assisted by Valmary, have a monopoly of the technical know-how, but the mechanical processing of the completed forms was taking place at INED. These academic pioneers of collaborative work with Henry effectively embraced a model of scientific and quantitative history that put into practice the injunctions of Ernest Labrousse and Fernand Braudel. More prosaically, in subsequent decades historical demography yielded an inexhaustible supply of topics for diplômes d’études supérieures and master’s theses (DH, 1980). By the late 1950s, however, voices were being raised among historians to denounce a method so time-consuming and repetitive that it deprived students of effective training in the discipline. The students themselves were not the last to be aware of these shortcomings [55].

49It is hard to gauge the extent to which these reactions drove Louis Henry to lead historical demography into a second, more ambitious phase. Between 1953 and the late 1950s, his main goal was to test a method and tools designed for data extraction [56], as well as to compile results from different geographical areas. But, in 1958, Henry changed the scale of his undertaking by presenting the project of a national survey in the journal Population. The aim was no longer to obtain results of general demographic value, but to reconstitute the population of ancien régime France by means of a purposeful and systematic survey. Seventeen years later, the survey provided the basis for a special issue of Population still regarded today as the basic reference on the demography of France in the modern period (Population, special issue, 1975; Dupâquier, 1988, p. 4; Brian, 1994, p. 9-24; Séguy, 2001). For Henry, history had become an objective in itself, and historical demography, which was overcoming the last pockets of resistance among historians [57], had established itself as a dominion of demography in historical territory [58].

1 – Natural fertility and reproductive biology

51The beginning of an answer lies in the uses of historical demography. Its implementation by Henry closely followed the development of the notion of natural fertility. This synchronism is, of course, anything but fortuitous. The demography of the past was regarded as an observatory whose findings could be applied immediately to decipher contemporary situations. The historical survey allowed a benchmarking of all the results obtained for the physiological determinants of fertility, and could therefore be used simultaneously for demographic and biological purposes.

52The study of the variation of sterility as a function of age provides an example. Louis Henry estimated it by combining conclusions based on “statistical” populations making little use of contraception (nineteenth-century England, rural Japan between the wars) and on “pre-statistical” non-contracepting populations (Geneva and Henripin’s French Canada). The convergence of results made it possible, in particular, to measure the age at which the rise in sterility accelerates [60]. Henry also formulated proposals concerning the nature and extent of “adolescent sterility” [61]. Likewise, he recognized early on that an increase in the last two birth intervals in the birth history of couples was characteristic of natural fertility. Through the accumulation of results, he was able to calculate the effect of infant mortality on the change in the length of the intervals (Henry, 1958).

53With time, the comparisons between non-contracepting populations, or between women in a given population, showed the importance of the variability of fertility in a “natural” regime. This dual variability increased the importance of breast-feeding, to which Henry was consistently attentive (Henry, 1952, 1957 and 1958). In all the cases examined, historical demography served as an adjunct science to biology. Henry’s findings were taken into consideration along with those concerning the Hutterites, a North American religious sect that did not use contraception. Although small and sociologically distinctive, it possessed “modern” statistical records. As Henry himself admitted, the consistency between the findings of Eaton and Mayer (1953) and their successors and those obtained by Henry enhanced the credibility of his results [62].

2 – Time and the other [63]

54Historical demography also had more indirect uses. Natural fertility was supposed to characterize non-contracepting populations in the developing world [64]. After coming up against the gaps in statistics on the French colonies early in his career as demographer, Henry attributed the behaviour of preindustrial Europeans to the populations under French rule [65]. The ambiguity of the concept of natural fertility was highlighted by these space-time comparisons: after a few years, Henry concluded that fertility levels were subject to continent-specific variations, a finding that reinforced his view on the “socio-cultural” underpinnings of natural fertility [66].

55For the time being, natural fertility made it possible to calibrate the results obtained in specific contexts, to situate populations with respect to one another, and to detect any implausible results that might be due to an idiosyncrasy of the data source [67]. This became a privileged area of application for historical demography. Another application was in the area of methodology. Louis Henry transposed the experience acquired in his monographs to help demographers, and even colonial administrators, design their surveys in the French colonial empire or in developing countries [68]. At one point, Henry even nurtured the ambition to make international organizations finance the development of historical demography [69]. His recommendations were transmitted all the way to the UN Population Division [70]. They focused, in particular, on sampling techniques and the checking of results. From this standpoint, it would be hard to overstate the contribution of historical demography and the “parish” approach to the changes then under way in the discipline. For a long time, demography and population science had worked at the level of aggregate populations, documented by national, regional or municipal statistics. The concept of the quantitative survey that focused in particular on recent advances in sampling methods (Blondiaux, 1998) and on the use of small samples, was ushering in a new methodological universe. Henry articulated the distinctive features of this type of “statistics of small numbers” by comparison with the classical techniques of demography [71].