Severity and Selectivity of the Black Death and Recur- ring Plague in the Southern Netherlands (1349-1450)

The Black Death is the textbook villain when it comes to the study of historical diseases and to the general public it remains a thought-provoking subject. To illustrate, in 2017 over three million viewers accessed the English Wikipedia’s Black Death page, compared to present-day Ebola which only had less than one million. Despite the wide drawing power of the Black Death, some of its most basic characteristics are still debated in academic circles. The focus of this paper will be on the severity of the Black Death and recurring plague outbreaks in the Southern Netherlands. More specifically it will reflect on the general assumption that plague evolved from a ‘universal killer’ to a more selective and less severe disease over time. Due to the scarcity of late medieval sources and a lack of quantifiable indicators, little is known about the causal mechanisms at work during the late Middle Ages. This paper offers a newly-compiled database of 25,610 individuals that died between 1349-1450 in the County of Hainaut to test a number of assumptions on the selectivity and severity of late medieval


Introduction
Plague is considered the most deadly infectious disease in history. Caused by the bacterium Yersinia pestis, it sparked off three widespread pandemics: the Justinian plague of 541-767, the second pandemic which started with the Black Death in 1347, 1 and the third pandemic which occurred during the middle of the nineteenth century. Although less severe than the two previous pandemics, with mortality mostly affecting India, it spread plague to ports across the globe. 2 Despite being perceived as a historical disease, plague is still present in wild animal reservoirs around the world and human plague-casualties are frequently reported. 3 In recent years, the increasing number of plague resurgences has led the World Health Organization to categorize the disease as a re-emerging global health threat. 4 Scholars have voiced concerns that future climate change may increase plague activity, 5 and some even consider it a potential bioterrorism agent. 6 In order to understand the causal mechanisms behind the spread, occurrence and severity of the disease, research is increasingly oriented towards historical plague in the hope that a long-term analysis will provide insights not attained by studies of contemporary (isolated) outbreaks. 7 At the same time there is growing consensus that, in order to assess the socio-economic and demographic implications of diseases over time, debates on the nature of epidemics and the changing characteristics of pathogens also need to be taken into consideration. 8 Even though we cannot assume pre-industrial epidemiological experiences mirrored the modern one, 9 especially given the disparities between the second and the third plague pandemics, 10 a greater understanding of the biggest killer-disease ever to afflict mankind is fundamental to predict the dynamics of future large-scale epidemics.
Most research on late medieval plague has primarily focused on the initial outbreak of the disease, constituting a kind of myopia in which interest in the Black Death seems almost perennial while late medieval echo-epidemics have received far less attention. 11 Due to the scarcity of sources we know little about the characteristics of recurring plague outbreaks in the century following the Black Death. 12 As a result, a number of assumptions have gained foothold in the international literature without a solid empirical foundation. To highlight this problem, this article will reflect on the state of current knowledge regarding the severity of the Black Death and the role of selective mortality in later plague outbreaks to explain their reduced mortality rates (section 2). 13 As indicated, one of the biggest hurdles is the scarcity and disparate nature of available sources on epidemiological characteristics. In order to understand the impact this has had on our understanding of late medieval plague epidemics, this article introduces data from a single continuous source, the Hainaut mortmain accounts (section 3). Using a database of 25,610 individuals that died in the County of Hainaut between 1349-1450 this article will analyse severity (in the context of acquired immunity), geographic selectivity and gender selectivity during the Black Death and recurring plague waves. The results of this analysis will be contrasted with certain beliefs in current literature (section 4). To conclude, I will discuss the need to question assumptions in the absence of hard data when studying historical plague outbreaks (section 5).

Black Death mortality
The Black Death caused demographic carnage of unprecedented magnitude, to which scholars have allocated a prime role in: regulating the equilibrium between population and resources, 14 determining the path of economic development, 15 the creation of institutions, 16 and both the little-and the great divergence. 17 Ascertaining the overall mortality rate has therefore attracted a fair amount of scholarly attention. Although absolute mortality from the 'Spanish' influenza pandemic of 1918-1919 was higher, 18 expressed as a percentage of the population, the Black Death was the biggest killer disease ever to have struck Europe, spreading to the Middle East and Northern Africa as well. Whilst earlier studies -relying on limited data -estimated overall mortality at 20 per cent to 35 per cent, current estimates have been pushed up to around half the European population. In fact, ongoing research is continuing to revise estimates upwards, a trend also seen in studies on early modern plague. 19 The highest mortality estimate for the Black Death is 60 per cent and was advanced by Ole Benedictow in his seminal book, The Black Death: The Complete History. 20 Even though it may become the new standard estimate, there are some issues with how it was calculated. First, for the purpose of his argument, Benedictow tended to avoid data that contradicted his story of exceptionally high mortality. 21 Moreover, in calculating an estimate for the whole of Europe, he noted a remarkable similarity of mortality levels in different regions. 22 However, it is well established that the Black Death did not hit all regions with the same severity. 23 Finally, Benedictow's claim that the data is sufficiently widespread and numerous to allow for a Europe-wide estimate, needs to be nuanced. 24 In truth there is an absolute scarcity of documentary evidence to furnish quantifiable indicators for mortality during the late Middle Ages, making it impossible to specify the exact numerical impact of the Black Death in a satisfactory manner. Even if sources are available, estimates remain inexact due to a lack of detailed accounts of deaths occurring during the time of plague. 25 Instead, mortality rates are usually calculated on the basis of anecdotes, chronicles and random samples of general economic data. 26 Despite the challenges posed by historical sources, plague research has witnessed some remarkable breakthroughs in the past two decades, most importantly with the identification of Yersinia pestis as the causal agent of the Black Death. 27 In recent years, aDNA testing and genome sequencing has allowed micro-biologists and bio-archaeologists to study plague on a geographical scale that often dwarfs localized studies of the past. 28 The allure of this wide geographic scope has enticed scholars in the social sciences (often non-historians) to perform studies on a pan-European scale as well. They usually do so not by relying on evidence gathered in the field or in the laboratory, but by using pre-existing historical datasets. When employing this data, some scholars tend to loosen the rigorous standards of evidence and interpretation demanded in the field of history, undermining the reliability of their results. 29 At the core of this problem is the assumption that Black Death mortality estimates are hard data suitable for complex analysis. In what follows, I will illustrate that certain datasets provide a poor starting point for comparing the severity of the Black Death with that of later plague outbreaks. This problem is compounded by the fact that there is currently no systematic overview of mortality rates caused by many of the echo-epidemics. In fact, even datasets tracking plague occurrences geographically over time have proven to be problematic. 30 It is assumed the Black Death acted as a 'universal killer' causing it to be more severe than subsequent plague outbreaks because (among others): it hit an immunologically naïve population, was less selective in who it killed and was more territorially pervasive. 31 Recurring plague outbreaks, on the other hand, are said to have been more selective. However, our understanding of how these epidemiological characteristics evolved and how they influenced severity is limited. 32 The Hainaut mortmain database provides an excellent opportunity to unveil some of the characteristics behind the severity and selectivity of late medieval plague. Before turning to these issues let us turn to the starting point, the mortality estimates for the Black Death in the Southern Netherlands.
One of the most widely used datasets of Black Death mortality rates, building on pre-existing research, was compiled in the book, Interdisciplinary public health reasoning and epidemic modelling: The case of the Black Death. 33 Based on their data, George Christakos and co-authors argued that 'unlike Northern Italy, (Black Death) mortality in Belgium was the lowest among present-day countries, with about 20 per cent mortality at most' . 34 I will test the reliability of the data on which this claim is based as a way to reflect on the need to adhere to strict standards of interpretation when using historical sources.
First, let us examine the possibility of comparing on a 'countryby-country' level to substantiate the claim that Belgium witnessed the lowest mortality rate of all present-day countries. To calculate the mortality estimate for an entire country, the authors were forced to extrapolate results from localised studies. This methodology has to contend with the fact that mortality rates could differ noticeably from one locality to the next. 35 For example, work with manorial sources for England and fiscal sources for France and Italy has revealed local population losses ranging from 30 per cent to 70 per cent. 36 For countries with a relative abundance of sources, such as England and France, the cross-referencing of multiple local estimates might yield some degree of reliability when extrapolated to a 'country-level' total. However, for countries such Next, let us consider the claim that Black Death mortality in Belgium was 20 per cent at most. Since this figure was attained by calculating the average for eight separate estimates, the question becomes 'how reliable is each individual estimate'? Table 2 represents the data used by Christakos et al., indicating the location, the start-and end date of the Black Death, the estimated mortality and the corresponding source. For the cities of Bruges, 37 Brussels, Ghent and Ypres the mortality estimates were based on information from Robert Gottfried's contested book, The Black Death: Natural and human disaster in Medieval Europe, which stated that '(in) Ghent, Bruges, Ypres, Brussels, and Antwerp […] plague mortality was "only" about 20% to 25%, not much more than it had been during the great famines of the 1310s' . 38 These percentages were not calculated by Gottfried himself, instead they were taken from publications by Van Werveke and Blockmans. 39 The comparison with mortality during the Great Famine is problematic as Van Werveke stated that data was only available for Bruges and Ypres. Moreover, neither Van Werveke, nor Blockmans give explicit reference to a mortality rate for any of these cities (Brussels is never even mentioned!), and Blockmans specifically noted that 'we can bring little precision in the mortality figures' . 40 It is therefore surprising that the mortality rates proposed by Blockmans for Maubeuge and Ath, 24 per cent and 16 per cent respectively, are so exact. These figures were calculated on the basis of the Hainaut mortmain account and refer to an average death rate per month. However, Blockmans exclusively used data from the first extant account which only covers the period from 24 June 1349 to 11 April 1350, while Black Death mortality was present until the end of 1351 in the County of Hainaut. Moreover, the mortmain documents for the Black Death period show several issues causing it to underreport mortality (see following section).
For the city of Antwerp, the reference is to a book by Naphy and Spicer in which they argued that, 'many historians comment on the "mild" plague that struck Antwerp […] (which) "only" killed one out of every four or five persons' . 41 This claim is highly contentious as there is no reference to literature or archival sources to substantiate it.
The final area for which the mortality rate was quantified is the County of Flanders for which a mortality rate of 16 to 25 per cent was based on the following quote by David Nicholas, 'Flanders may have lost one-quarter to one-sixth of its population during the plague of 1349, which is less severe than most other areas' . 42 Nicholas referenced the work by Adriaan Verhulst, who in turn used publications by Sivéry, Blockmans and Van Werveke to attain these estimates. 43 As such, it is merely a summary of the estimates already presented for individual localities.
For the remaining cities of Louvain and Liège the mortality estimates were not quantified, instead Christakos et al. indicated 'no impact' by relying on an argumentum ex silentio, assuming that an absence of sources mentioning plague equalled an absence of plague activity. For Liège, a book by Scott and Duncan was used, referencing the claim that 'places such as Milan, Liège and Nuremberg, escaped the disease' , again without reference to literature or archival sources. 44 For Louvain, a book by Maria Kelly was referenced in which we find the unsubstantiated claim that, 'plague spared some crowded urban centres like Milan, Ghent and Louvain' . 45 All these claims are unconvincing, as Ole Benedictow has persuasively argued that absence of evidence should not be interpreted as evidence of absence, especially when dealing with scanty late medieval sources. 46 Finally, for the region of Brabant, where the impact was deemed low, an article by Despy was referenced in which the impact of the Black Death is described as mild. 47 This claim is based on a narrative source from the second half of the fifteenth century, 48 however, Despy also mentioned an earlier chronicle from the late fourteenth century which does indicate severe mortality during the Black Death. 49 Although it is beyond the scope of this article, it would be interesting to see how estimates for other countries (in the dataset) hold up to scrutiny. If the case study for Belgium is anything to go by, the basic data on the impact of the Black Death compiled by Christakos et al., is flawed to say the least. 50 However, this has done little to deter scholars from employing the data at face value for complex regression-and spatiotemporal analyses. 51 What this section has shown is that mortality estimates for the Black Death used in recent literature do not provide an adequate basis for comparing severity, both across regions and across time. This makes any general claim on the diminished severity of recurring plague outbreaks difficult to demonstrate empirically.

The mortmain database
In order to shed new light on the selectivity and severity of the Black Death and recurring plague outbreaks, I present findings from a database of 25,610 individuals found in the mortmain accounts of the County of Hainaut during the period 1349-1450. 52 The term 'mortmain' is reminiscent of the legal term found in regions with English, French and Anglo-Norman juridical authorities, referring to the alienation of land to the dead hand of the church. 53 However, the Hainaut mortmain accounts are more comparable to English heriot taxes since both were 48 Anecdota ex codicibus hagiographicis Johannis Gielemans (Brussels 1895). 49 Radulp De Rivo, Historia episcoporum Leodiensium (Liège 1616). 50 Much of the problem has to do with a lack of a standardized methodology to calculate mortality rates, which in turn is due to the disparate nature and scarcity of available sources. paid as a death duty in the form of the best movable possession of the deceased. The main difference is that heriots were levied on tenants payable to a manorial lord, while mortmain was levied on a broad range of people that came from an ancestral lineage of servitude.
By the fourteenth century only a small minority of the population in Hainaut were still serfs and mortmain no longer functioned as an exclusive sign of servile status. 54 As early as June 1349 (first extant document) the accounts offered a separate rubric that read 'received from serfs, bastards and foreigners' , highlighting their separate status from the 'wider' population paying mortmain (see illustration 1). 55 While heriots targeted a specific socio-economic subset of society, mostly (male)  tenant heads of households, the reach of the mortmain was wider including all men and women who had reached age of majority (and sporadically emancipated children). 56 The main criteria was that an individual came from an ancestral line of servility, for which the mortmain accounts recognized three distinct lineage groups. 57 These groups were (a) people whose ancestry came from a lineage of serfdom in the broad sense, (b) the so-called 'sainteurs' who initially stood under the patronage of an ecclesiastical institution, and (c) people who originally stood under the patronage of a lay lord, the so-called 'gens d'avouerie'. 58 The number of people subjected to mortmain payment in Hainaut was substantial. An estimated 10 to 16 per cent of total mortality was recorded in the rural localities that appeared in the mortmain accounts during the first half of the fifteenth century (see appendix). Considering the coverage of probate inventories for Ghent between 1349-1400 has been estimated at 10 per cent, and factoring in the wider geographical reach, the mortmain accounts offer a significant insight in mortality in Hainaut. 59 More important than the numerical coverage, however, is the fact that mortmain was levied on a wide range of socio-economic groups. Only three groups were specifically exempt from payment due to their socio-economic status, these were: (a) 'les gens d'origine franche' (form of noble status transferred through maternal lineage), (b) 'les seigneurs haut-justiciers' , and (c) religious professionals. 60 Despite the exclusion of certain groups with high social status, not all high-ranked individuals were omitted from the accounts, and titles such as 'sire' and 'mayeur de [x]' are found in the documents. The accounts also extended to the poorest in society, labelling them as 'pauvre' or 'mendiant' . The mortmain accounts also recorded a significant proportion of women (43.7 per cent of the entire database). When removing a number of localities that had specific stipulations excluding women, a sex ratio of mortality of 1.07:1 can be derived from the data, indicating slightly more men. 61 Overall, it would seem that the mortmain accounts are one of the most complete and continuous sources on late-medieval mortality. However, their reliability as a proxy for mortality crisis has been criticized, most notably by Ellen Kittell who examined these documents for the County of Flanders. 62 There are some general source limitations -such as delayed on incomplete registration in times of crisis. Yet due to their serial nature, the principal advantage of mortmain accounts is that deaths can be compared in known crisis years with 'normal' years. 63 However, Kittell has pointed out that the count of Flanders was increasingly able to subject more people to mortmain payment in the course of the late fourteenth century, exploiting a population weakened by persistent plagues and upsurges in conflict. It was only after 1380 that people were recorded in a standardized way, the rubric of the document itself began to change, and explicit reference to the precise item seized occurred.

Map 1 Map of the county of Hainaut in mid-fifteenth century with indication of district-level boundaries
This was not the case for Hainaut, both structure and terminology of the documents remained unchanged from the first extant document of 1349 all the way up to 1500. Moreover, there do not seem to be any structural increases in the amount of people recorded not linked to mortality crises identified in the literature. The issue instead was that the number of districts (and therefore localities) included in the account showed fluctuations over time (see appendix). In order to assess the mortality spikes in the data, it is therefore preferable to show the average mortality per locality instead of absolute mortality, as shown in figure 1 below. Apart from the Black Death, for which raised mortality can be seen in the period 1349-51, figure 1  Unfortunately, the mortmain accounts only show a minimum impression of Black Death mortality for several reasons. (a) The first extant account, starting in June 1349, is in poor physical condition, approximately 20 per cent of the content is missing (see illustration 2), (b) this first document only covers a period of ten months, while most other accounts span twelve months, (c) the accounts during the Black Death period contain second recordings of additional deaths for certain localities, something not seen in any of the other accounts up to 1450 and probably indicative of disruption in local administration which was (d) in all likelihood caught off guard by sudden hyper-mortality. 65 Despite these issues, the mortality peak during the Black Death is clearly visible for the County of Hainaut in figure 1.
The use of mortmain accounts is not a novel approach to the study of demographic history in the Southern Netherlands. The documents were first used to analyse the impact of recurring plague outbreaks in a ma thesis written by Wilfried Vangassen in 1952. 66 In 1965 French historian Gérard Sivéry used the mortmain accounts to prove that the Black Death had a more severe impact on the County of Hainaut than previously assumed. 67 The following year, Sivéry used the documents to ana-  lyse the severity of the plague outbreak of 1400-1401 in Hainaut, showing significant mortality. 68 Most notably, however, was Wim Blockmans' use of the accounts in his authoritative article on the social and economic effects of late medieval plague in the Low Countries. 69 In this article, Blockmans refuted the claims made by Van Werveke that the Black Death had not reached the central areas of the Low Countries. 70 A notion which, until that time, had been accepted in the international literature. 71 Instead of a miraculous escape from plague, Blockmans introduced the notion that the Black Death had been comparatively mild in the Low Countries, antithetical to its status as a densely populated, commercialized and highly urbanized region. However, there are several issues with the use of the mortmain accounts in previous scholarship that need to be addressed here. The article by Blockmans seems to rely mostly on the dataset compiled by Vangassen in 1952 which poses several problems. Key among them are the errors that occurred either due to the recording of double entries or the failure to record more than one person in a line. More surprising is the fact that neither Sivéry nor Blockmans incorporated the account of 1350-1351 in their research. 72 A rather sizeable gap in the data considering the Black Death was not a one-year event. Similarly, the article by Sivéry which focused on the 1400-1401 plague outbreak did not incorporate the account from September 1400 to September 1401. 73 Moreover, previous research never analysed the mortmain accounts as a continuous serial source and ignored its value for analysing urban and rural plague dynamics, or gender-relate mortality effects.

Selective mortality effects
Based on the data compiled from the mortmain accounts, it is possible to analyze several factors that have been used in the literature to explain the diminished severity of recurring plague waves. A first factor thought to have caused lower severity in recurring plague outbreaks is the notion that humans could build up immunity to plague over time. 74 The assumption is that, because the Black Death affected a 'virgin soil' population, it caused higher mortality than subsequent plague outbreaks. 75 However, the prevalent idea that populations could permanently acquire immunity to plague (or other pathogens) if part of that population survived a previous epidemic has been deemed misleading. 76 The assumption that the Black Death was more severe than plague outbreaks in the following hundred years is constrained by the fact that there is an absolute scarcity of quantitative evidence for much of this period. 77 Based on the Hainaut database it is difficult to provide a conclusive answer because the documents offer only a minimum estimate for Black Death mortality. However, there are indications of higher severity of certain recurring plague outbreaks, despite prolonged contact of local population with the pathogen. For example, judging by the results from figure 1, the outbreak of 1400-1401 seems to be much more severe than the outbreaks of 1368-1369 and 1380-1382. 78 A second factor used to explain diminished severity is linked to the notion that the population was so immunologically naïve that the Black Death killed indiscriminately. 79 This has been compared to research on recurring plague outbreaks which suggests a higher selectivity with regard to age and pre-plague health status. On the topic of age, bio-arche- ological evidence suggests that the Black Death was selective towards adults with mortality risk increasing with age. 80 For subsequent plague outbreaks, however, this selectivity seems to have gone the opposite direction by affecting children to a disproportionate degree. 81 In the case of England or Florence, the second plague outbreak of the early 1360s is often labelled the 'plague of children' . However, according to Ann Carmichael, this may simply be due to a large baby-boom generation which dominated the population at risk a decade after the Black Death. 82 Concerning pre-plague health status, research has suggested a connection between the Great Bovine pestilence, which resulted in the loss of normal dairy consumption for people born in this period, and the severity of the Black Death. 83 Bio-archaeological research has indicated that stress seen in late thirteenth-and early fourteenth-century Northwestern Europe may have created favorable conditions for a severe plague epidemic. 84 However, historical research on pre-industrial famines has not found a clear causal connection between famines and plague. 85 Whatever the possible selective effects of recurring plague might have been with regard to age and health, there is research that suggests that the Black Death also had a selective impact despite causing mortality on a truly massive scale. 86 Unfortunately, the Hainaut database offers no direct information on health status or age. However, the database does provide sex-disaggregated data for those who died so the potential difference in sex-selective mortality effects for the Black Death and recurring plague can be analyzed. 87 In 2009 Sharon DeWitte published a bio-archaeological study which found no conclusive sign that sex played a role in plague's potential discriminatory effect during the Black Death. 88 In a follow-up article, however, DeWitte offered a more nuanced hypothesis. Either the Black Death killed more men who were vulnerable due to prior physiological stress, or it killed more women who were otherwise healthy. 89 Two main limitations of previous research has limited the progress on this topic. First, sample sizes are usually small because bio-archaeologists are restricted to using a limited amount of skeletons from a limited number of burials sites. Historians on the other hand usually focus on the results from isolated case studies, making it difficult to discern broader patterns. Second there is the problem of ascertaining whether differences in mortality rates between sexes are caused by factors intrinsic to plague or the result of differential exposure. 90 In late medieval and early modern Europe, women started migrating to the cities to a greater degree than men who remained in the countryside to work. Thus higher female mortality may be due to the disease ravaging urban localities. 91 What is needed to attain reliable results is data that can be compared over a long timespans between plague-years and 'normal'-years, as well as data beyond individual burial sites to compare sex ratios in mortality between rural and urban environments. This is where the Hainaut database offers a unique insight as it provides sex disaggregated data over the course of a century for both countryside and city. 92 Figure 2 indicates the main plague periods from 1349-1450, the peaks in mortality they caused and the sex ratios (%). The figure shows that both the Black Death and most of the recurring plagues during the late Middle Ages had a tendency to kill more women than normal. This is represented by the fact that when a major plague occurred, there was an inverse relationship with the sex ratio in mortality. The notion that the Black Death killed indiscriminately and that this effect changed in later plague outbreaks does not seem to hold for Hainaut, at least with regard to sex-selectivity.
A third and final factor is rooted in the assumption that the Black Death was a universal killer, hitting city and countryside in equal measure. Recurring plague, on the other hand, is seen as more selective and therefore less territorially pervasive, 93 with outbreaks often confined locally or regionally. 94 What happened in the period immediately after the Black Death is still largely unknown, 95 although according to Naphy and Spicer, 'the second outbreak and subsequent plagues differed in that it was more confined to urban areas' . 96 This assumption is  in line with a strong tendency of historians of European plagues to focus on data-rich environments, remaining unimpressed by patchy and problem-ridden evidence from rural hinterlands. 97 This has led to the common claim that plague became an almost exclusively urban phenomenon in (Southern) Europe by the fifteenth century. 98 Once again it seems that much of this assumption is based on the scarcity of data to systematically assert the 'universal' nature of the Black Death in contrast to later plague outbreaks. Using the Hainaut database to compare plague severity between rural and urban localities, it is possible to test this assumption. However, before turning to the analysis of the data, it  Figure 3 separates urban and rural mortmain data. In order to accommodate for different aggregate mortmain figures between city and countryside and changing population levels across time, urban and rural data were separated by using an average increase/decrease in deaths compared to the moving average of previous years. Previous averages are calculated by skipping back a year and calculating the average of the five previous years (excluding highest and lowest figures). The mortmain figure in the year in question is then divided by the average of the previous years, multiplied by 100 with 100 subtracted. For the Black Death, no previous documents were available. However, the urban aggregate mortmain comes very close to that of the rural one (209 to 228) and it is possible that the Black Death had a greater impact on the cities than the countryside. Figure 3 shows that many of the key-medieval plagues had discernible mortality effects on the countryside, with trends similar to the cities.

Conclusion
Current knowledge regarding the severity and selectivity of late medieval plague is still limited, mostly due to the scarcity and disparate na- ture of available sources seen throughout Europe. It is clear that certain datasets on Black Death mortality rates 'which have been widely used for spatiotemporal-and regression analysis' are unfit (for this purpose). A general revision of all pre-existing datasets by carefully checking mortality estimates against the actual content of available sources is necessary to ensure the continued and reliable contribution of historical data to the field of plague studies. However, this endeavour should go beyond reconstructing mortality rates for the Black Death, in fact, each widespread plague outbreaks that followed is deserving of the same amount of scholarly attention dedicated to the initial fourteenth century outbreak. This article has shown how a comparative study of plague severity, either across regions or across time, still has to contend with incomplete and flawed data.
Based on the Hainaut mortmain database, there are no noticeable changes in the selectivity of plague between 1349 and 1450. As such, certain assumptions on selective mortality effects can be refuted. In short, the narrative on selective mortality effects and diminished severity is not as black-and-white as some scholars argue. One possible explanation is that plague only became a 'selective' disease from the late fifteenth century onwards. 101 The assumed selectivity in mortality effects proposed by certain authors for the late medieval period might have been influenced by results for early modern plagues.
However, the findings of this article also have broader implications. In order to understand the diverging directions of socio-economic and demographic development seen after plague outbreaks, 102 it is vital to take into consideration the changing epidemiological characteristics of the disease over time. 103 It is therefore important to move beyond cemented assumptions to allow historical investigation to engage more fully with recent trends in plague ecology. With other sciences such as micro-biology and bio-archaeology becoming fully mature in their study of historical diseases, it is up to historians to insure their data(reliability) and methods mature at an equal rate. Future research will undoubtedly seek to incorporate the historical data, as the lack of longterm analysis is one of the major gaps in current disaster studies. 104  out, this explains (in part) the fluctuations in the number of districts. The most common pattern is that plague years not only cause higher mortality in localities that are ever-present in the accounts, it also causes low-population (or localities where the population subject to mortmain is low) localities to appear in greater numbers. These are localities that, in 'normal' mortality years, do not show a death in every single account, hence they are not ever-present. The increase in the number of (especially rural) localities during plague years has a direct impact on the representation of plague severity in figure 1. Because this figure calculates mortality over the number of localities, the actual severity may be underestimated. This is the result of the high number of localities that are present in these years that are not ever-present in 'normal' mortality years. The total number of deaths are therefore divided by a much larger group of localities. However, the increase in localities is in itself a sign of widespread mortality and thus a fac-  1349  1353  1357  1361  1365  1369  1373  1377  1381  1385  1389  1393  1397  1401  1405  1409  1413  1417  1421  1425  1429  1433  1437  1441  1445     tor of plague epidemics. In any case, it does not influence the conclusion made in section 4 that the 1400-1401 plague outbreak was more severe than several previous plague waves. The number of localities in 1400-1401 is higher than at any point in the fourteenth century and still the severity is noticeably elevated.

Estimates coverage of mortmain accounts
In an effort to analyse the importance of the Hainaut mortmain accounts relative to the total population in the County of Hainaut, I present an estimation of the coverage. In correspondence with the available hearth counts, a coverage has been calculated for 1406 and 1444. 106 The calculations below are based on a number of assumptions which will be explained more in-depth in the final section of this appendix. Most important are the average size of hearths (expressed as a multiplier) and the average mortality during 'normal' years (expressed as a percentage of total population).
Tables 3 and 4 use the following method to calculate the coverage.
(1) First they take the absolute number of hearths for those localities that appear both in the mortmain account and in the hearth count.
(2) Next, a multiplier ranging from 3.5 to 5 is applied to calculated the total population represented by the hearth counts for these localities.
(3) Based on these results we attain a number that indicates the 'estimated total population' of those localities. These differ according to the average hearth size (multiplier) that has been applied. (4) The following step is to calculated the average mortality using 'normal' mortality rates of 3, 3.8 and 4 per cent. (5) This results in the number of people that likely died during the year in which the hearth count was performed based on the assumption that it was a year showing average mortality. (6) Based on the mortmain accounts, a five year average is calculated of the people that died in the selected localities, the corresponding hearth count year is used as middle value for this calculation. This has been done to insure that any outliers specific to one single year are removed from the calculations, thereby providing an estimate of the average mortality. (7) This result is compared to the total number of deaths attained by the previous calculation to arrive at a percentage that indicates the likely coverage of the mortmain accounts in the recorded localities. The calculation is represented schematically below.
Mortmain 5y avg (HearthCount * multiplier) * avg mortality % = est coverage % Based on the calculations, the overall coverage of the mortmain is very similar for 1406 and 1444, between 10 and 16 per cent. However, the sample is average size of a hearth tended to fluctuate quiet considerably during the late Middle Ages. 115 For example, Jan Dumolyn claims that after a period of scarcity, famine and epidemic disease the multiplier should be lowered. 116 This is why the calculated coverage of the mortmain accounts used multipliers ranging from 3.5 to 5 when estimating the total population.

Mortality rate
The tables below give a brief overview of the mortality rates that have been proposed in the literature for the Southern Netherlands during the late medieval period. Because the hearth counts for Hainaut only contain rural localities, the choice was made to use the mortality rates for rural settings.