Why a long-term perspective is beneficial for demographers

Although many contemporary demographers pay attention to historical demography, there is often a surprising lack of appreciation of the demographic circumstances and systems of the past, suggesting an implicit assumption that they are not relevant to the present or that the methods, data, and questions addressed by historical and contemporary demographers are different. This paper provides an overview of historical demography as published in Population Studies and how this has developed over time. Drawing on this, I demonstrate that historical and contemporary demography use similar data sources and identical methods, and they often address comparable questions. I argue that an appreciation of demographic patterns and processes is beneficial for all demographers, even those who work on the most recent time periods, and that better integration of historical and contemporary demography would be beneficial to both. The paper also considers three challenges for historical demography as it moves forward.


Introduction
Population Studies is a journal that has always aimed to encompass a broad view of demography (Grebenik 1996), and like many demography journals, it includes a range of papers devoted to historical topics. Its small editorial board has frequently included a historical demographer, with previous consulting editors including Roger Schofield and Bob Woods. The 50th anniversary special issue of Population Studies included a contribution on historical demography (Saito 1996), and the British Society for Population Studies annual conference always includes a lively historical demography strand. While this identification of historical demography as a subdiscipline is much welcomed, and could be argued to valorize the work of historical demographers by sending the message that historical topics are valued and their authors will feel at home, it could alternatively be thought to ghettoize historical demography, labelling it as something which is not relevant or of only tangential interest to 'proper' demographers. In this paper I demonstrate that historical demography uses similar data and the identical methodological toolkit to more current demography, and it addresses some of the same questions and issues. I argue that the work of both non-historical and historical demographers can benefit from a long-term perspective and that a failure to appreciate historical findings can lead to misinterpretation. Although I consider that continuing to define historical demography as a subdiscipline is useful, more needs to be done to 'mainstream' historical demography.
I start by discussing how historical demography can be defined and providing the definition of it that I use in this paper. I then illustrate the breadth of topics covered by historical demography. I do not want to repeat other recent and excellent overviews of the recent history of historical demography.
[These contributions include: Antoinette Fauve-Chamoux's edited volume, A Global History of Historical Demography, which contains chapters detailing the development of historical demography in individual countries or regions, as well as an introduction to international networks and interdisciplinarity in historical demography (Fauve-Chamoux et al. 2016); a special issue of Annales de Démographie Historique to honour the 20th anniversary of the French Society for Historical Demography, which contains papers on a variety of topics frequently addressed by historical demographers (Boudjaaba et al. 2015); George Alter's examination of the derivation and importance of methods in historical demography since the 1950s (Alter 2019); and Myron Gutmann and Emily Klancher Merchant's recent overview of the origins, methods, and findings of the subfield in the Handbook of Population (Gutmann and Merchant 2019).] For this reason, and because this paper appears in the 75th anniversary issue of Population Studies, I concentrate my discussion on historical demography as published in this journal. Drawing on this, I then consider whether there are important differences-in terms of data, methods, and topics-between historical and other branches of demography. The following section argues that a long-term perspective is valuable for all demographers, and the penultimate section considers three challenges for historical demographers: the changing data universe, interaction with other disciplines, and decolonization.

A working definition of historical demography
If we are to compare historical demography with non-historical demography (henceforth referred to as contemporary demography), it is important to be clear about what is meant by historical demography and conversely what is meant by contemporary demography. It is easy to identify a topic or paper as relating to fertility, mortality, or migration, and many historical demography papers are also easy to identify. Those which relate purely to previous centuries and to extinct demographic regimes or events can unproblematically be labelled as historical. However, there are several grey areas which make a neat definition of historical demography difficult. The most obvious of these relates to the location of the dividing line between 'history' and 'the present'. Some historians study very recent periods, and according to a survey in the BBC History Magazine conducted in 2015, anything over 10 years in the past can be counted as history (History Extra n.d.), but this definition is not helpful for demography. Even the most recently available demographic data are often not particularly freshthere is normally a period of several years between the collection of census or other large survey responses and their availability for academic analysis, so the most up-to-date data are often well over a decade old. According to this definition, all demography except projections, forecasts, and nowcasts would be historical.
Historians themselves draw no clear dividing line between the past and 'current affairs'. Some identify a particular year as distinguishing the present from history, but this tends to be related to the context of the place they are in and the type of history they study; for example, for historians of modern Russia, an obvious watershed year might be the collapse of the Soviet Union in 1991, although increasingly the transition from Soviet to post-Soviet eras raises fascinating questions of continuity and change for historians (Kritika editorial 2010). In a demographic context, contenders for such landmarks could include the first demographic transition or the baby boom of the 1950s and 1960s. However, this is also problematic: the first demographic transition was completed over a century ago in Europe but is still ongoing in some African countries. The baby boom offers a fixed time point but arguably provides a boundary which is less relevant to many parts of the world (although see Reher and Requena (2015) for an argument that the baby boom was a global phenomenon). Fixed time points are also problematic as they make it difficult to classify papers written at different points in time. The baby boom is arguably history today, but it was not history for papers written in the 1960s, and the use of fixed time points leads either to anachronistic categorization or to a very distant location of any dividing line between history and the present. The fact that a given time period or event becomes history with the passing of time is simply the nature of history and is not intrinsically problematic, but when conducting a survey of historical material over time, it is inappropriate to use fixed time points.
Another approach might be to distinguish historical demography papers by their use of particular types of source material, along the lines drawn by John Goldthorpe in the early 1990s, when he suggested that the fundamental difference between history and sociology lies in the nature of the evidence and how it comes into being (Goldthorpe 1991). He argued that historical evidence is always incomplete and finite and not usually designed for the research uses to which it is put. Sociologists, however, can design surveys to collect precisely the data they need to answer their research questions. Along these lines, Demographic and Health Surveys (DHS) might be regarded as a modern source but parish registers as historical. Goldthorpe went on to argue that the sparseness of historical evidence means that historians need to arm themselves S158 Alice Reid with particular techniques, and this might imply that historical demographers use different methods and possess different skills from contemporary demographers, an unsatisfactory suggestion which I return to later in this paper. Goldthorpe's notion that historical sources are never systematic or scientifically designed, whereas modern sources always are, is also contradicted by reality: on one hand historical censuses were carefully designed and their coverage is excellent; on the other hand there is increasing use of 'found' data, such as data from digital traces. This reveals an intrinsic flaw in Goldthorpe's argument, as it destroys what could be seen as a fundamental relationship between history and the past. Even if Goldthorpe were right about the nature of sources, with his definition (as for an event-based one) it is difficult to update the location between history and the present with the passing of time.
An alternative approach, which makes it easier to compare historical papers published over a long time spectrum, is to identify as historical those relating to a given number of years before publication. In this approach the number of years used will always be quite arbitrary. Steven Ruggles used 30 years in his analysis of historical papers (Ruggles 2021), but given that demographers frequently look at time series, I have defined historical work as that which presents data and analysis primarily relating to over 50 years before the year of publication. While this solves some problems, there is still a grey area relating to papers which cross this dividing line, for example those covering a period of 70 or 100 or so years, coming right up to the present: a recent example is the paper by Martin Dribe and Christopher Smith, who examined social class differences in Swedish fertility between 1922 and 2015 (Dribe and Smith 2021). In the interests of inclusivity, I have defined such papers as historical because a substantial part of their analysis relates to a time over 50 years before publication, but I do not mean to imply that they cannot also be contemporary. It is less obvious how to categorize those where the period of interest only just enters my definition of historical and the analysis mainly relates to a more recent time, for example Jane Falkingham and Arjan Gjonça's paper on 'Fertility transition in communist Albania, 1950-90' (Falkingham andGjonça 2001). The authors of these papers may not think of themselves as writing a 'historical' paper; nevertheless, for the sake of consistency and the need for a clean cutoff, I have included the few such papers in my definition of historical papers.
Another definitional issue relates to papers that assess intergenerational patterns or the influences of early-life effects on later-life outcomes. The outcomes in such papers may occur in the present or very recent past, but the influences often relate to well over 50 years in the past. Are papers which examine the effect of childhood disease or famine on later-life mortality, for example, contemporary (because of the temporal location of their outcomes) or historical (because of the temporal nature of the influences)? Sometimes such work is done by 'contemporary' scholars and sometimes by those who specialize in a historical time period, and the focus of the papers may vary. There are arguments for saying such papers (like those that cross the temporal dividing line) are both historical and contemporary, and I have included them in my consideration of historical papers in Population Studies.
For the analysis described in the following section I used titles, abstracts, and skim-reading of papers where necessary to identify 371 historical papers published in Population Studies from volume 1 in 1947 to papers published (or available online) as of 1 July 2021. (Keywords were not used because they were only introduced in 2003, because relatively few historical papers identified themselves in this way, and because the word 'history' can throw up red herrings in the form of event history analysis, work history, fertility history, and so on.) I included papers where at least some of the events of interest took place more than 50 years before the publication of the paper. I excluded the majority of papers analysing fertility among cohorts born 50-70 years before publication, on the basis that most of the events of interest (births) did not occur longer ago than 50 years before publication (I mention this because dates in the titles of such papers often referred to the cohorts defined by years of birth). However, I retained papers looking at later-life effects of early-life conditions on the basis that the early-life conditions were also of analytical importance. I excluded replies and responses to other papers, as these rarely included analysis and were often very short, and I also excluded the majority of review papers and most of the personal reflection papers appearing in the 50th anniversary issue of 1996 and in Population-The Long View (Coleman et al. 2015). Nevertheless, this is arguably a more inclusive definition than that used by Melinda Mills and Charles Rahal in their analysis of papers in Population Studies in this volume (Mills and Rahal 2021); their Historical demography category included rather fewer papers. My discussion of Population Studies papers also occasionally refers to parallel or diverging trends in historical demography published in other outlets, although these are not based on a systematic classification. In the following sections I contrast these with contemporary demography papers, which are defined (in contrast to historical demography) as papers where the events of interest took place within the 50 years prior to the date of publication.

Historical demography in the pages of Population Studies
The birth of historical demography is usually located in the 1950s with Louis Henry's popularization of the method of family reconstitution: that is, the reconstruction of life histories by linking the baptisms, marriages, and burials recorded in parish registers (Wrigley 1969, p. 81;Rosental 2003;Gutmann and Merchant 2019). The baby boom had unexpectedly reversed a sustained fertility decline, shaking demographers' confidence in their ability to forecast future trends (Hobcraft 1996), and Henry hoped that better knowledge about past demographic regimes and changes would aid this situation (Rosental 2003). However, he was not the first to examine demographic factors in historical populations, nor did he invent family reconstitution (Schlumbohm 2018), and this pre-existing demographic interest in historical topics is evident in the first two issues of Population Studies, which were commissioned by David Glass (Grebenik 1996). These issues contained papers on the French demography of the eighteenth century (Vincent 1947) and a historical review of the vital statistics of Norway (Backer 1947). These papers demonstrate what was to become an enduring commitment to historical topics in the pages of Population Studies. Figure 1 shows the number and percentage of papers published in Population Studies over time that can be considered to be historical demography (according to my definition). While most periods are decades, the first covers only four years, and the last includes a small number of papers published in 2021 (or published online and not yet assigned to an issue). There have been many influences on these trends, including developments in the discipline, new sources of data, popular issues, and the expansion of publication outlets, as well as differences in these factors between contemporary and historical demography. Nevertheless, the graph clearly identifies that the heyday of historical demography occurred between the 1970s and the 1990s. This is most pronounced in terms of the number of papers, as the fact that this period also saw the most publications in the journal in total means that the variation in percentage of papers is rather less pronounced. Across the whole period around one in five papers in the journal can be considered historical.
Most overviews of historical demography over time suggest that parish registers (as a source) and family reconstitution (as a method) dominated the discipline until the 1970s (Saito 1996;Gutmann and Merchant 2019). In some sense Population Studies was no exception to this. The growth in historical papers in the journal until its peak in the 1970s reflects the growth and peak in parish register papers, of which 22 were published that decade. These covered topics including data quality (e.g. Levine 1976), methods (e.g. Sharlin 1978), mortality shocks (Hollingsworth and Hollingsworth 1971), population growth (Norton 1971), household structure (Bradley and Mendels 1978), nuptiality (Martin 1977), and bastardy (Laslett and Oosterveen 1973), as well as the subject which family reconstitution is particularly well suited to: the many aspects of fertility (too many papers to single any out).
However, it would be a mistake to assume that historical demography was limited to the analysis of parish registers. Figure 2 shows the percentages of historical papers published in Population Studies that used primarily individual-level, aggregatelevel, or 'other' data (papers based on discussion of other people's individual-level results were classed as aggregate-level). Even in the heyday of parish register analysis, papers using individual-level data (including parish register based studies) were a minority, and individual-level data sources (such as parish registers) were generally used to illustrate community-level patterns and processes. In this era, historical demography-like demography more broadly, as noted by Ridhi Kashyap in her contribution to this volume (Kashyap 2021)-was dominated by aggregate-level sources and populationlevel processes.
During the last 25 years, there has been renewed growth in the use of individual-level data and also a diversification of individual-level sources, as noted in Gutmann and Merchant (2019). These sources include continuous registration data kept by Swedish churches (e.g. Dribe 2004;Edvinsson et al. 2005) and state population registers in Belgium and the Netherlands (e.g. Bras and Neven 2007;Steenhof and Liefbroer 2008). In some cases, these can be linked to other data sources to add rich contextual data and informative covariates, opening up the possibility for more precise examination of the influences on the different experiences of different groups. Prospective longitudinal data S160 Alice Reid are rare in historical demography, but some authors have exploited health surveillance records on young children to examine patterns and differentials in health and survival. Foundling hospitals have proved a particularly good source for Italy and Spain (Kertzer et al. 1999;Revuelta-Eugercios 2013), and London's foundling hospital-the only one in the UK-has also enabled interesting insights (Levene 2005). Other health surveillance data include records kept by UK health visitors in the early twentieth century (Reid 2001(Reid , 2002 and insurance records (Riley 2003). Anthropometric databoth on heights, derived from army recruits or militia lists (Öberg 2015;Jaadla et al. 2021) and on body mass index extracted from other sources (Murray and Lagger 2001)-have been used as indirect indicators of cumulated net nutrition, although the complexities of interpreting such data as indicators of diet or disease have also been pointed out (Alter 2004).
The recent growth of census microdata, facilitated by large-scale digitization projects and coordinated collections, has started to bear fruit in the pages of In Figure 2, the 'other' category mainly covers two sorts of papers: the first set of papers analysed demographic writing and population thought and were published mainly before 2000 (e.g. Hutchinson 1959;Spengler 1970;Van Praag 1978). Papers in the second group used a variety of qualitative material to examine birth control, abortion, contraception, and attitudes of ordinary people or authorities (Blacker 1957;Peel 1963;Sauer 1974Sauer , 1978Santow 1995;Van de Walle 2000). Although papers in both these groups have been published by the journal throughout its existence, the last 25 years has seen a decrease; nevertheless, such papers continue to appear in journals specializing in qualitative and archival research [for a recent example, see Schult (2020)]. This might be due to a positive choice on the part of the authors, but the spread of an idea that papers in demography journals must be quantitative in nature might also have deterred them from offering papers to such outlets.
These changes in data sources were accompanied by changes in the focus on different historical periods addressed by papers in Population Studies. Figure 3 shows a heat map of the percentage of historical papers published in each decade that The benefit of a long-term perspective S161 addressed different historical periods. Papers addressing more than one century have been allocated to each century they covered, so percentages may add up to more than 100. Darker colours in many of the cells relating to papers published before 1951 indicate that such papers were particularly likely to cover a long time frame. The nineteenth century has always received considerable attention, but more revealingly, the graph shows that the last two decades has seen a reduction in papers relating to earlier periods. This is largely due to the decline in parish-register-based papers and the growth in those using census microdata and other individual-level sources.
As noted earlier, the impetus for the development of historical demography was the need to understand of fertility patterns and trends in pre-transitional and transitional populations, and parish reconstitution  Note: Percentages may add up to more than 100 as some papers with a long time span may be assigned to more than one century. Source: As for Figure 1. S162 Alice Reid was particularly well suited for the analysis of fertility. Fertility might therefore be expected to have dominated historical demography papers, and Gutmann and Merchant (2019) stated that there has been less historical demography work on mortality than fertility, but this does not seem to have been the case in Population Studies. Figure 4 shows the percentages of papers published in each decade that focused on fertility, mortality, migration, and nuptiality, respectively (note that a paper could be classed as focusing on more than one of these, so percentages add up to more than 100). The percentages focused on none of these four processes are also provided; this category encompasses papers on demographic thought, population size and structure, sources, household structure, morbidity, and other aspects of health. Overall, 38 per cent of historical papers in Population Studies focused on fertility and 43 per cent on mortality. There were only two decades, the 1970s and the 2010s, when fertility was a more popular topic than mortality.
Historical fertility research published in Population Studies has always covered a wide variety of topics. Alongside papers establishing the levels and trends in fertility, common themes have been age patterns, fecundity, premarital pregnancy, and relationships between fertility and marriage, particularly for the pre-transition period. These papers have gone far further than simple description, and there are several contested issues that have been repeatedly addressed. For example, although early research on pre-transitional populations using parish registers established that couples were unlikely to be deliberately limiting the size of their families (Henry 1961), there has been an ongoing debate about whether they engaged in deliberate spacing behaviour (Knodel 1978;Van Bavel 2004). Recent approaches to this topic have investigated the timing of the fertility response to short-term economic shocks and argued that temporarily reduced fertility was the result of deliberate postponement of the next birth rather than involuntary mechanisms such as spousal separation, maternal illness, or foetal loss (Bengtsson and Dribe 2006;Jennings et al. 2017).
The progress and process of the decline in fertility during the first demographic transition has received considerable attention, and the debate, initiated by the publications of the Princeton European Fertility Project, about whether this was due primarily to changes in the costs and benefits of children or to cultural or ideational changes has been a major preoccupation (e.g. Brown and Guinnane 2002). The relative contributions of parity-specific limitation (which theorists of natural fertility suggested was dominant), of spacing, and of non-parity-specific limitation has been another unresolved issue which continues to be addressed on the pages of Population Studies and elsewhere (Hionidou 1998;Gortfelder and Puur 2019). Some authors have used the sex composition of offspring to generate insights into this topic (Sandström and Vikström 2015;Gortfelder and Puur 2020). More recently, identification of the social groups in the vanguard of fertility decline and investigation of how ideas about fertility control spread have been facilitated by the availability of new individual-level data and the use of multilevel and spatial analysis (Dribe et al. 2017;Dribe and Smith 2021). For example, Sandra González-Bailón and Tommy Murphy used agent-based modelling to investigate the role of social interactions and religion on the diffusion of smaller family norms (González-Bailón and Murphy 2013). As the passing of time has shifted the baby boom into the category of historical demography, this has also featured in recent papers on historical fertility in the journal (Gauvreau et al. 2018; Van Bavel et al. 2018;Reher and Requena 2020).
The topic of mortality in historical demography is also home to a number of unresolved debates. Probably the most prominent of these relates to the causes of mortality decline and the relative contributions of nutrition vs public health, a debate provoked by Thomas McKeown over 50 years ago in a series of papers in this journal (McKeown and Record 1962;McKeown et al. 1975). Most of the direct responses to McKeown have been published elsewhere, but other examinations of the relative impacts of nutrition, public health, and medicine on mortality have been examined in this journal (Guha 1993;Jaadla and Puur 2016). Infant mortality itself has dominated papers on mortality since the early 1990s, with two out of every five historical mortality papers since 1996 addressing early-age mortality and investigating aspects such as social and spatial differentials, religion, infant feeding, water supply, maternal nutrition, the clustering of infant deaths in families, and often a combination of these.
Crisis mortality has always figured highly: the mortality effects of famine, infectious diseases (smallpox, cholera, plague, and influenza), and war have been frequent topics. These have ranged from new estimates of the extent of a mortality crisis (e.g. Chandra (2013) on 1918 influenza in Indonesia, and Vallin et al. (2002) on famine and war in midtwentieth-century Ukraine), through identification of the structure of mortality (e.g. Hoch (1998) on age and cause-of-death patterns in Russian The benefit of a long-term perspective S163 subsistence crises of the nineteenth century; Hionidou (2002) on cause of death in Greek famines; and Alfani and Bonetti (2019) on age, sex, and social status differences in plague mortality in seventeenth-century Italy), to questions of whether exposure to famine in utero or infancy has a scarring or selection effect on mortality in the long term (Doblhammer et al. 2013). Other papers have broadened the perspective beyond the mortality effects of epidemics, looking at morbidity (Benedictow 1987), the effectiveness of disease control initiatives (Meegama 1979), policy among colonial authorities (Ambirajan 1976), and the effects on property structures (Alfani 2010). Of course, crises often affect other demographic processes too, and the complexity of the demographic responses to subsistence crises (including fertility and/or migration responses) have been explored in contributions from Pedersen (1995), Song (2013),  among others.
Given the fact that the method of parish reconstitution revolves around a couple's marriage, it is perhaps unsurprising that the popularity of nuptiality as a topic coincided with the peak in papers using that method. An economically responsive age at marriage and neolocal household formation were established by early historical demographers as two of the defining characteristics of the (Northwest) EMP and as key mechanisms for the maintenance of moderately low fertility (for a pretransitional population) and a high-wage economy through the process of Malthusian homeostasis (Hajnal 1965;Laslett 1969;Wrigley 1978). The existence and form of the EMP in North-West Europe are broadly unchallenged, but this does not mean it is well understood. Although it has received little attention in Population Studies (possibly because the main focus is on economic growth, and marriage is just used as an indicator), there is a lively debate in the economic history literature about the role of the EMP and (relatively) high autonomy among women in driving economic growth, the Little Divergence (the development of greater prosperity in the North Sea area than in Southern and Eastern Europe), and the Great Divergence (when Western European economic growth started to outstrip that in Asia) (see De Moor and van Zanden (2010) and Dennison and Ogilvie (2014) for two particularly influential papers). In this debate, as more generally, the EMP is often presented as having consistently high ages at marriage and levels of celibacy, with some researchers failing to appreciate the way that marriage responded to wages and led to long, selfcorrecting cycles of economic and demographic growth before the industrial revolution.
As noted by Ronald Skeldon in his contribution to this volume (Skeldon 2021), migration is a neglected topic in demographic journals in general, and historical demography within Population Studies is no exception. Such migration work as has been published has involved estimating the volumes and characteristics of international migrants from

S164 Alice Reid
Europe to America or Australia (e.g. Eurenius 2020), the effects of migration on population structure in migrant-sending nations (e.g. Murphy 2016), and the effects of migration on the calculation of demographic indicators (e.g. Wrigley 1994).
Finally in this overview section, it is worth considering the geographic balance of historical demography papers in Population Studies. Nearly 40 per cent of historical papers published in the journal since 1947 have related to the British Isles, although the imbalance is reducing and the figure was only 15 per cent in the most recent decade. The majority of the remaining papers have related to Europe, with Sweden, Germany, and the Netherlands featuring highly. The United States has been well represented, as has Asia, from which papers have mainly related to China, India, and Japan. There have been very few papers from Latin America and the Caribbean and even fewer from Africa. (I did not investigate the geographic balance of contemporary papers in Population Studies, but Mills and Rahal's (2021) analysis in this volume indicates that although there is still a European bias when all papers are considered-with around 35 per cent of papers (where Mills and Rahal were able to ascertain the geography) using European data-this is less concentrated than for historical papers alone). This geographical imbalance might have arisen if historical demographers were primarily interested in demographic transition, as there has been more time to study places with earlier demographic transitions. However, it has already been demonstrated that historical demographers are not simply interested in demographic transition, but in demographic variation and the influences on it: examples of studies which have examined these outside Europe and the Anglophone world include Vanlandingham and Hirschman (2001) and Drixler and Kok (2016). It is likely that data availability and accessibility have played a considerable role in this uneven geographical coverage. Some countries, for example China and Japan, benefit from rich historical data in the form of prospectively collected genealogies, although the fact that these consist of surviving patrilineages introduces biases that need to be carefully considered in analysis (Zhao 2001). Some such studies have not reached the pages of journals because the highly successful Eurasia Population and Family History Project concentrated on book production (Bengtsson et al. 2004;Tsuya et al. 2010;Lundh and Kurosu 2014). However, Population Studies has attracted a few papers on the more recent historical demography of China and other Asian countries; these used later surveys to examine changes since the mid-twentieth century or intergenerational effects of famine (Song 2013;Babiarz et al. 2015;Jennings 2017). Some of the Anglophone bias might be a product of the fact that France, Spain/Portugal, and Italy all have long-established historical demography societies with associated historical demography journals (Annales de Démographie Historique, Boletin de la Asociación de Demografía Histórica, and Popolazione e Storia), and the first two of these in particular may attract offerings from the wider francophone and hispanophone worlds. In contrast there is no long-established specific historical demography journal in English (although The History of the Family increasingly concentrates on demographic topics and a new journal, Historical Life Course Studies, was established in 2014). I return to the Eurocentric nature of historical demography later in the paper.

How different is historical demography?
Having briefly reviewed the topics covered by historical demography, particularly in the pages of Population Studies, I return to the definition of historical demography and specifically examine what, if anything (apart from time period), makes it different to the rest of demography.

Data
The data used by historical and contemporary demographers are largely the same: censuses (both in aggregate and microdata forms), vital event data (births, marriages, deaths, and moves), and longitudinal studies. However, there are differences. Historical demography has so far rarely used sample surveys, at least not those designed for the collection of demographic data, including the World Fertility Survey (WFS), DHS, and other household surveys. However, the life histories produced by family reconstitution can be analysed in similar ways to the fertility histories generated by the WFS and DHS. Parish registers are perhaps uniquely used by historical demographers: in essence they are equivalent to civil registration, but issues of under-registration of infants who died before baptism, patchy survival, inconsistent recording of information, indecipherable handwriting, small populations, small name pools, and migration mean that they require more consideration of coverage, completeness, and bias. This is not to say that contemporary demographers do not pay attention to such issues (Ní The benefit of a long-term perspective S165 Bhrolcháin et al. 2011), and anyone using longitudinal data must grapple with loss-to-follow-up bias. However, the incompleteness of the life histories produced by family reconstitution means that the set of rules that needs to be followed in order to avoid informative censoring is complex and restrictive (Alter et al. 2020). The proportion of historical demographers now using family reconstitution is rather small, and such issues are arguably less relevant to those who use other data. However, even with careful treatment, the perspectives offered by different types of data source can differ, and this is a point historical demographers (and others) need to constantly remind themselves of (see van den Berg et al. (2021) for an enlightening comparison of linked registration data and population registers).

Methods
Historical and contemporary demographers also use the same toolkit. The classic demographic methods (e.g. life table analysis), newer demographic methods including decomposition (e.g. Torres et al. 2019), and a full range of regression techniques (including Cox proportional hazards, logistic, and Poisson regression) are all used across the discipline. Advances in regression-such as multilevel modelling to take account of the nested nature of much demographic data and spatial regression to account for geographic dependency-have been embraced by historical demographers (e.g. Szołtysek et al. 2021). Perhaps the most recent addition to the regression armoury is cure regression, which can separate out the eventual chance of an event from its timing (e.g. Gortfelder and Puur 2020). Various forms of simulation, including agent-based modelling, have been used to assess and correct biases in data (e.g. Jonker and van der Vaart 2007), understand outcomes (e.g. Zhao 2001), and test hypotheses (e.g. González-Bailón and Murphy 2013). Methods developed for use in 'less developed' countries to generate mortality and fertility rates from census responses have proved very useful for the late nineteenth and early twentieth centuries in settings (e.g. the UK) where census microdata are plentiful but individual event data are scarce. These methods include the indirect estimation of child mortality (Haines 1995) and, for fertility, reverse survival (Spoorenberg 2019) and the Own Children Method (Reid et al. 2020).
Specific methods for historical demography do exist, but these are generally still based on core demographic principles (Alter 2019). For example, family reconstitution is similar to other linkage methods, and back projection-used to estimate population totals from counts of events-can be seen as a form of simulation. SMAMs (singulate mean ages at marriage) and Coale's indices of fertility and marriage are based on life tables, stable population theory, and demographic standardization (Alter 2019).

Topics
The brief overview earlier demonstrated that historical demography addresses a far wider spectrum of topics than the 'war, famine, and the first demographic transition' definition provided as a helpful indicator of the sort of topics covered in Mills and Rahal's (2021) Historical category (although I'm sure they did not mean this to exclude other topics). It is true that war and famine are more likely to be covered in historical than contemporary demography, but crises are not peculiar to the past, as Covid-19-and the research it is inspiring-is demonstrating (see Mamelund and Dimka (2021) in this volume). Neither is the first demographic transition a purely historical phenomenon, and issues to do with the second demographic transition and its relationship to the first and to the structure of pre-transitional societies have occupied many historical demographers, who are often deeply interested in continuities and changes over the long term and between historical periods (Duranton et al. 2009;Reher 2021). Many of the big questions preoccupying historical demographers are still pertinent today, including (but by no means limited to) the ways that fertility decline manifests in terms of parity (Hruschka et al. 2019).
Of course, there are differences in the questions and the balance of topics. Within historical mortality, infant mortality is particularly prominent, partly due to the dominance of this age group within mortality as a whole, but also because migration means that family reconstitution yields smaller quantities of often less representative data on mortality at older ages. Although similar factors also traditionally yielded larger amounts of work on early-age mortality (compared with older-age mortality) in lowand middle-income countries, the balance is shifting as age structures change (Preston 1996;Van Raalte 2021). Ageing as a topic has so far rarely been addressed for historical populations, again mainly because populations did not start to age appreciably until after the first demographic transition. However, a handful of papers in Population Studies S166 Alice Reid have provided estimates of the numbers of very old and their mortality (Thatcher 1992;Garson 1991) and the effect of fertility fluctuations on ageing and its consequences (Rowland 1984;Reher 2015).
Marriage is arguably a more important topic in historical demography than contemporary demography, given the key role of marriage in maintaining a low-pressure demographic system in North-West Europe (Hajnal 1965;Wrigley 1978). The role of decreasing age at marriage in fuelling the baby boom (Van Bavel and Reher 2013) and the importance of age at marriage as part of legacy family systems that are arguably continuing to dominate geographic variation in fertility patterns (Reher 1998) mean that marriage is likely to be of continued interest to historical demographers. In many parts of the world, however, marriage is no longer a good marker of union formation, and so it is less likely to be studied by contemporary demographers.
While there are differences, then, in terms of data, methods, and topics, the similarities between historical demography and contemporary demography are arguably far greater than between other subdisciplines within demography. As mentioned earlier, there are also distinctly grey areas, in the form of research that considers long-term trends that end near the present, that compares historical and contemporary situations, and that examines the influence of early-life conditions on later demographic outcomes. This all suggests that treating historical demography as a subdiscipline might be unhelpful, as I elucidate in the next section.

Intradisciplinary communication and the value of a long-term perspective
I suggested in the 'Introduction' that separating historical demography from the rest of demography could signal that it is not relevant, or of only tangential interest, to contemporary demographers. If this is true, then it is likely that many contemporary demographers have little conception of the longer time picture of the patterns and processes in their area of interest. It may also be true that historical demographers have little appreciation of the present day. Perhaps this should not be thought of as a problem: maybe we should not worry if scholars of mortality do not know much about fertility. But one of the strengths of demography is that it pays attention to the links between demographic processes, and most demographers would agree that at least some understanding of multiple processes is important (in contrast to, for example, epidemiologists, who focus on health and mortality). It is similarly, or maybe even more, worrying if those investigating fertility in the present, for example, have little understanding of its long-term patterns, or those studying past fertility are unaware of the latest trends.
Penelope Corfield has argued that history is important because it 'helps people to establish a secure footing or "location" within the unfolding saga of time', and this is also true of academic research (Corfield 2008). The present or recent past, and indeed the future, are part of a long continuum of continuity and change-both evolution and revolution-and it is very difficult to make sense of a single moment or short segment of this without some appreciation of the long term. Demographic time series often start no further back than the mid-twentieth century, and this is problematic because for many places, including the UK, the 1940s to 1960s were highly atypical in terms of family formation. More young women got married in those decades than at any time since the sixteenth century, and they did so at younger ages. Together with the recuperation of births delayed by the second world war, this led to unprecedently high period fertility. Perhaps not unrelatedly, the mid-twentieth century was the pinnacle of the 'traditional' male breadwinner family, at least in the UK. The evidence suggests that in pre-industrial times it was common for married women to engage in wage-earning or 'productive' work (not just domestic labour), but such opportunities decreased steadily over the late nineteenth and early twentieth centuries. Using the mid-twentieth century as a benchmark creates an impression that the situation then was the historical norm, and this may exaggerate change over time in measures such as period fertility that are affected by timing issues. Although consistent data sources often do not go back further than the mid-twentieth century-for example, in Britain, age and parity of mother began to be collected at birth registration only from 1938-analogous data for earlier periods often exist. An example of a publication making an effort to place findings in a longer-term context is Berrington and Simpson (2016), while Schürer et al. (2018) is primarily a historical piece that also includes contemporary comparisons. This is not to say that all demographers should be historical demographers or that a long time period should always be used. However, we should all broaden our perspectives so that we are aware when there is something relevant to consider.
Another example of an apparent lack of historical perspective can be seen in the series of papers The benefit of a long-term perspective S167 initiated by Link and Phelan (1995) about the fundamental causes of disease. Socio-economic status differences have held perennial interest for demographers, and research has arguably been boosted by Link and Phelan's fundamental cause theory. This contends that whatever the proximate causes of health gaps within populations-such as differential access to curative or preventive medicine, or differences in sanitation, water supply, nutrition or knowledge-these are themselves caused by social status, making social status the fundamental cause of health inequality (Link and Phelan 1995). One of their most recent papers on the topic, published in the journal Demography (Clouston et al. 2016), has provided a more historically sensitive argument than earlier versions, recognizing that as medical knowledge and material circumstances changed, social forces would affect different diseases at different times. However, the central assumption that social status differences in health have been the norm throughout human history remains, and the authors did not engage with the literature from historical demography which suggests that social differences in health were not always present but emerged in the seventeenth century and fluctuated over time (Antonovsky 1967;Woods and Williams 1995). Given their disciplinary backgrounds, it is perhaps understandable that the authors were not aware of this literature, but it is more surprising that their attention was not drawn to it by reviewers who, given the journal, are more likely to have been demographers.
The lack of a long-term perspective can lead to a tendency to 'read history sideways', a phrase coined by Thornton (2001). This practice involves taking cross-sectional observations from contemporary societies, assuming these represent different stages of a universal developmental paradigm, and combining this developmental trajectory with patchy historical evidence to fill gaps in the historical record. Arland Thornton argued that this practice can lead to a misconception of the history of family change and fertility decline, and to developmental idealism which furthers ethnocentric ideas about the desirability of 'western' family and societal forms (Thornton 2001). Historical evidence demonstrates that the family forms and demographic structures of the European past rarely map onto those seen in late-twentieth-or early-twenty-first-century 'less developed' countries, so a better knowledge of history can be an important corrective to the tendency to read history sideways.
The very welcome development of specialist meetings, such as the European Society for Historical Demography (ESHD) biennial conference, may have had an unintended consequence of making dialogue between historical and contemporary demographers more difficult, in that it may have siphoned off research that would otherwise have been presented at general demography conferences. Matthijs et al. (2016, p. 14) made the point that this apparent marginalization of historical demography was the natural consequence of the maturation of the discipline, but they also suggested that opportunities to contribute to key debates from a historical perspective were being lost.
Of course there are many demographers, both historical and otherwise, who show evidence of routinely taking a broad perspective, and this is aided by papers that consider theories about process or change in relation to both the past and present while at the same time paying attention to particular contexts. Many of the most influential scholars in the field have published such papers: examples in Population Studies relating to theories of fertility change include papers by John Knodel (1977), Jack Caldwell (1981), and John Cleland and Christopher Wilson (Cleland and Wilson 1987). More recently David Reher (1998Reher ( , 2021 has taken a different but equally informative approach, arguing that current European patterns in household and family formation are largely determined by long-standing family systems, with a 'weak' family system in North-West Europe and a 'strong' one in the South. More papers of this nature would benefit the integration of the discipline, as a recent commentary argued: 'a renewed focus on big questions, rich description and historical narrative might even draw to our field the wider attention it so richly deserves' (Drixler and McCants 2016, p. 120). In addition, it would be beneficial for historical demographers to consider explicitly the ways in which their work might be relevant to contemporary demographers and to take opportunities to enter into dialogue.

Three more challenges for historical demography
Some of the short contributions in the collection of short papers produced and distributed at the second conference of the ESHD in 2016 (Matthijs et al. 2016) express a worry that historical demography is under threat. With the demise of pure demography master's courses and a reduction in student funding, many contemporary demographers feel the same way, and it is difficult to know if the S168 Alice Reid threat to historical demography is any more severe. However, the humanities, particularly history, are in danger in many British universities, with the UK government pushing for a focus on science, technology, engineering, and mathematics (STEM) and vocational courses, which it perceives to be of 'higher value' (Fazackerley 2021). There is also a perception that the Research Councils, which are the major sources of funding, are narrowing their remit: the Arts and Humanities Research Council steering clear of quantitative history and the Economic and Social Research Council insisting on policy relevance. The next section of the paper discusses three challenges which the historical demography community needs to confront as it moves forward.
The implications of a changing data universe Traditionally historical demographers have had to construct their data themselves, including transcribing them or overseeing local transcription and sometimes undertaking a time-consuming process of data linkage (by hand or machine-assisted). That lengthy process often generated a deep engagement with and knowledge of the data, including consideration of who created the data in the first place and the circumstances in which they did so, as well as issues of coverage and completeness, how inconsistencies in recording and difficulty in reading might introduce inaccuracies, and possible ways these could be addressed. Such processes allow a deep understanding of the data and their context and can greatly aid the data cleaning and coding, the creation of proxy variables, the analytic strategy, and the interpretation. Boulton and Davenport (2015) provided a good example of the way that understanding the changing contexts of death registration can avoid misinterpretation of a lengthening interval between birth and baptism.
The model of researchers creating their own data sets, however, is becoming outdated. Transcription of new 'big data', such as full-count census microdata, is often subcontracted to firms in countries where labour costs are low, or sources can be transcribed using online crowd-sourced labour, with double entry to ensure accuracy. Handwritten text recognition (e.g. transkribus.eu) is also becoming increasingly feasible. While these are very welcome advances in data capture (facilitating the production of more publications, a clear advantage in today's 'publish or perish' culture), any deep knowledge of the data must be gained consciously rather than soaked up through experience. The different relationship between the researcher and the data may contribute to the increasing dominance of advanced statistical methods in the discipline and make it harder to become attuned to the data's historical context (Kesztenbaum 2016). As Anders Brändström lamented: 'lost are the stories' (Brändström 2016, p. 151). This is not to say that previous generations of historical demographers were paragons of good practice in this respect. John Hobcraft wrote in 1996 that 'we spend too little time trying to explain and to understand, rather than to quantify and to describe the past' and argued that we need more qualitative analysis: 'more cogitation and weighing of fragmentary evidence and less datadredging … without abandoning our undoubted quantitative strengths' (Hobcraft 1996, p. 488). This is still true today, and the changes in data and computing power mean that it is even more important to warn users that variables in cleaned and harmonized data sets may not have straightforward interpretations that match up with their variable names and descriptions. This is a particular challenge when using big data, especially for researchers focusing on a subset of a larger data set, as even when they do make an effort to understand the data, in-depth investigation into local contexts may be simply impossible. This can be partially overcome if those writing metadata and user guides ensure that these are comprehensive and detailed, and data users will need to pay particular care in reading these documents and must consider issues of interpretation as they analyse the data. These issues may be particularly important to new users of demographic data from outside the discipline of demography, as explained in the next subsection.
Disciplinary turns: The qualitative turn in social science, the quantitative turn in history, and the historical turn in economics In the late twentieth century, the 'qualitative turn', also known as the 'cultural turn', led to a move away from quantitative analyses in the social sciences and history. Although historical demography did not lose faith in quantitative methods, Steven Ruggles (2021) has argued that there was a parallel 'historical turn' among the more quantitative social sciences such as economics, and that the last decade also saw a revival of quantitative history. Economists and economic historians are becoming increasingly aware of the potential in The benefit of a long-term perspective S169 newly available big data sets (e.g. full-count census microdata), the increased accessibility of existing data sets deposited in public online archives, and the possibility of scraping data from the internet. They are keen to apply sophisticated econometric analyses to address issues of change and causation, and although many economists and economic historians do pay careful attention to censoring and coverage (and some of the most careful historical demography is carried out in economic history departments, e.g. the lively historical demography group within the Centre for Economic Demography at Lund University in Sweden), this is not always the case. Some economists without demographic training appear to believe that statistical methods can overcome biased data (e.g. Cinnirella et al. 2017). George Alter, Gill Newton, and Jim Oeppen have recently demonstrated that ignoring family reconstitution rules designed to avoid 'informative censoring'-where a period of observation ends with the event of interest-results in seriously biased estimates of fertility rates (Alter et al. 2020).
Herbert Klein (2017) has suggested that it is only historians (by which he appears to include demographic historians) who have the patience to immerse themselves in a data set to such an extent that they develop a deep and nuanced understanding of its provenance and interpretation. While there may or may not be a distinction between demographic historians and historical demographers, the two groups are certainly alike in their attention to the detail of the data, and many feel threatened by the possibility of economists taking shortcuts and producing erroneous results but appearing more efficient. More interdisciplinary communication is perhaps one way to address this, requiring a willingness from both economists and demographers to meet and listen: in the UK one welcome step in this direction is an annual historical demography workshop, established a few years ago by economic historians at LSE.

Representation, diversification, and decolonization
I have already demonstrated that historical demography papers published in Population Studies were most likely to be about the UK, Europe, and wealthy countries in the anglophone world, and this raises issues of representation and decolonization. Former British colonies, with linguistic and institutional colonial legacies, are probably more accessible to anglophone scholars, hence their better representation in Population Studies. There is certainly a paucity of papers relating to the historical demography of South America in this journal, even though there are historical demographers in this region (e.g. at the University of Campinas in Brazil), and parish registers for a number of parishes in Brazil and Peru have been scanned and digitized by the British Library's Endangered Archives Programme (eap.bl.uk). It is likely that such scholars publish in Spanish or Portuguese, although those using recently available census data deposited at IPUMS may publish in English (e.g. Castro Torres 2021). Cordell's (2000) overview of African historical demography indicated that around half the material was published in French and also that little ended up in demography journals. It is also worth noting that much of the African historical demography covered by Cordell related to the late twentieth century, calling into question the relevance of my arbitrary cut-off for inclusion as historical.
One of the major reasons for the lack of historical demography studies in many of today's less wealthy countries is data availability. Sweden serves as a benchmark for historical populations, primarily because it has particularly detailed data available from an early stage. In contrast, many countries, including most of those in Africa, don't carry out regular censuses or have complete-coverage vital registration for the present, let alone for the past. Surviving records were almost all created by colonial infrastructure, whether evangelizing missionaries or government surveys (Notkola et al. 2000;Walters 2016;Meier zu Selhausen et al. 2018). Some of these relate only to the colonizing population (Cilliers and Mariotti 2021), so while certainly valuable and informative, they can only yield information about an atypical subset of society and not about the Indigenous population. Those which do relate to the local population need careful interpretation in terms of both coverage and content, as well as consideration of the biases conferred by colonial control and surveillance of the population (Walters 2016). While more parish registers likely exist and could be exploited, fluidity in naming and large gaps between events and recording, among other issues, mean that data linkage with these sorts of missionary registers will never yield data as complete as those provided by historical European registers. Perhaps, however, we should not see this paucity of data as a problem and instead embrace Christopher Clapham's (2020) exhortation for a release from Northern structures of knowledge creation as a necessary part of a decolonization project. It might be unnerving for historical demographers to let go of their S170 Alice Reid tools and approaches, but we should remember that unconventional sources and qualitative approaches have always been part of the historical demographer's armoury. It is also difficult to envisage the forms and comparability of knowledge that will emerge, but Shane Doyle (2000Doyle ( , 2008 has demonstrated that, for some places at least, other surviving records, oral histories, and ethnographic data can provide valuable insights. An embracing of alternative data and a qualitative attention to interpretation may reveal new understandings to counter and complement the narratives offered from the perspective of colonial legacy data.
Perhaps more worrying from a decolonization perspective is the fact that African historical demography has to date been undertaken largely by white scholars from privileged backgrounds. This might be connected to research priorities and funding availability, which push historical work down the research agenda in African universities, and it is difficult to know how to respond to this particular challenge.

Concluding thoughts
This paper has provided a brief overview of papers published in Population Studies that can broadly be identified as historical demography. This analysis has illustrated the breadth of historical demography research and the similarities and differences between historical and contemporary demography. I have argued that a long-term perspective can be immensely valuable to both historical and contemporary demography. For contemporary demographers, I have tried to demonstrate how difficult it is to understand the present without some knowledge of the past. A lack of this knowledge can lead to an assumption that today's circumstances, or those of the midtwentieth century when many reliable data series began, are some sort of historical norm. Alternatively, it can lead to the assumption that societal forms in today's poorer countries also occurred at 'earlier stages of development' in the pasts of today's rich countries (the practice of reading history sideways). Historical demography has often provided an important corrective to such problematic assumptions and has thereby contributed to demonstrating the situatedness of knowledge production in demography.
I have suggested that categorizing papers as historical demography is not always helpful and can contribute to a growing feeling that historical research has little relevance for contemporary demography; this diminishes the quality of insights into both past and present. Readers might think that a recommendation for historical demography to be jettisoned as a subcategory of demography will naturally follow, but in many instances it can be helpful and instructive for historical demography papers and researchers to be identified as such, particularly when no other single category is appropriate.
Calls for more attention to a historical perspective are not new: in 1996 John Hobcraft, himself echoing John Hajnal and William Brass, wrote 'we must pay at least as much attention to learning from and analysing the past as to the more speculative aspects of future trends' (Hobcraft 1996, p. 488). However, more still needs to be done to mainstream historical demography, and better communication is key to this. Contemporary demographers could try to attend historical conference sessions and historical demographers likewise attend contemporary sessions, but attention to each other's papers might be easier to engineer through more sessions that include both historical and contemporary perspectives on the same subject. Historical demographers could write more explicitly about the relevance of their research for contemporary demography, in order to persuade contemporary demographers to pay attention to a broad time perspective. Some papers could revolve around 'big picture' questions, but there is room for such points in the introductions to many more standard papers. Communication is also key to the first two challenges mentioned. The challenge of a changing data universe can be at least partially met by detailed and comprehensive documentation on the part of data creators and careful attention to this by data users. Better communication with those entering historical demography from outside the field would encourage a more sensitive appreciation of demographic issues. Decolonization offers a different, and perhaps more difficult, challenge, but meeting it is likely to involve identifying new forms of information, embracing new methods, and paying careful attention to the way that these are culturally produced.

Notes and acknowledgements
1 Please direct all correspondence to Alice Reid, Department of Geography, University of Cambridge, Downing Place, Cambridge, CB2 3EN, UK; or by Email: amr1001@cam.ac.uk or Twitter: @amrcampop 2 Acknowledgements: Thank you to two enthusiastic anonymous reviewers, to Wendy Sigle and Rebecca