Urban sanitation and the decline of mortality

ABSTRACT

This paper introduces a special issue of The History of the Family on sanitation and urban mortality. The special issue contains papers which focus on the impact of sanitary reforms on mortality change in Australia, Switzerland, Finland, Sweden, and England and Wales. The current paper outlines the main features of the debate over the causes of mortality change and the role played by sanitary reforms in this. It then highlights some of the methodological and other challenges posed by the definition of ‘urban’ areas, the identification of relevant sanitary reforms, and the choice of dependent variables. The paper then proceeds to summarise the main features of the individual papers before drawing some conclusions for future research.

KEYWORDS:

• Sanitation
• public health
• mortality

The history of mortality decline continues to play a central role in economic, social, medical and demographic debates. Whilst one line of enquiry has continued to examine the role played by diet and nutrition alongside other factors (see e.g. Floud, Fogel, Harris, & Hong, 2011; Fogel, 2004), other authors have paid renewed attention to the impact of environmental factors, especially in urban areas. A number of writers have focused on the impact of atmospheric pollution (see e.g. Bailey, Hatton, & Inwood, 2018; Beach & Hanlon, 2018), whilst others have continued to highlight the importance of sanitary advances (e.g. Chapman, 2019; Kesztenbaum & Rosenthal, 2017). However, the precise extent of the contributions made by different factors continues to be debated.

One of the main areas for discussion concerns the extent to which existing work has continued to focus on the decline of mortality in a relatively small number of locations and countries. The contributors to this special issue seek to address this question by introducing a number of new case studies. Although three of the papers continue to focus on England and Wales, the special issue also contains papers on Australia, Finland, Sweden and Switzerland. Although most of the papers focus directly on urban locations, they also look at rural areas and examine the process of mortality change in urban areas of various sizes.

The question of urbanisation has long been central to studies of mortality change. In 1955, the sociologist, Kingsley Davis (1955, p. 433), estimated that the proportion of the world’s population living in cities containing more than 20,000 inhabitants had increased from 2.4 per cent to 20.9 per cent between 1800 and 1950, and this figure has continued to increase since that point. In 2017, the World Bank estimated that almost 55 per cent of the world’s population could be classified as urban-dwellers.¹ However, for much of human history, towns and cities have also been regarded as demographic ‘sinks’, with especially high rates of premature death and excess mortality (Dyson, 2011). Consequently, the decline of urban mortality is a particularly important part of the overall history of mortality change.

The papers in this special issue are especially concerned with this topic and with the role played by sanitary intervention within this. Taken together, they aim to shed new light on the role played by sanitary intervention in different parts of Europe (including England and Wales) and Australia during the late-nineteenth and early-twentieth centuries. These years have often been regarded as critical to the process of demographic and epidemiological transition within these countries (Caldwell, 2006; Omran, 1971). They also witnessed the origins of a substantial shift in the relative healthiness of urban and rural areas, by the end of which the healthiness of many urban populations had surpassed that of their rural counterparts (Leon, 2007).

In addition to its intrinsic importance, the history of mortality decline has also received attention from writers interested in the role played by demographic changes in the process of economic growth. The decline of infant and child mortality enabled more people to survive to ‘productive’ ages, and increases in longevity at older ages have also yielded a substantial ‘demographic dividend’ (Lee & Mason, 2006). Improvements in physical health and cognitive performance have further enhanced economic productivity (Floud et al., 2011).

This introductory paper sets the scene for the special issue. We begin by providing a brief summary of the historiography of mortality decline in the countries under review. We then explore some of the methodological and definitional problems associated with the study of urbanisation and its implications for the study of urban mortality. Subsequent sections look at the broad history of sanitary intervention before proceeding to introduce the papers themselves. The paper concludes by summarising the implications of the individual case-studies for future research.

1. The decline of mortality

The interpretation of the causes of mortality decline has long been contentious. During the first half of the twentieth century, British historians and economists often argued that a large part of the decline should be credited to improvements in medical provision. For example, John Plumb (1950, p. 78) claimed that the decline in the death rate after circa 1740 was ‘almost entirely due to improved midwifery … and the foundation of lying-in hospitals’, whilst John Hicks (1942, p. 43) attributed it ‘beyond all doubt’ to ‘the improvements in sanitation and medical skill which were beginning to be effective in northern Europe by the middle of the eighteenth century’.² However, although some writers have recently sought to revive the claims made on behalf of medical intervention, most observers now believe that the primary responsibility for improvements in mortality lies with some combination of improvements in diet and nutrition, housing, and sanitary intervention (see e.g. Johansson, 2010; Van Poppel et al., 2016).

One of the most important contributors to these debates was Thomas McKeown, Professor of Social Medicine at the University of Birmingham from 1945–77.³ He argued that the decline in mortality in England began during the second half of the eighteenth century and was largely associated with a decline in the number of deaths associated with infectious diseases. He also identified a number of factors which might have been responsible for this decline, including a decline in the virulence of infectious organisms, increases in human immunity, medical improvements, improvements in sanitation, and improvements in diet and nutrition. He argued that none of the first four explanations could really account for either the timing or nature of mortality change, with the result that the only remaining candidate was an improvement in nutrition (McKeown, 1976; McKeown & Brown, 1955; McKeown & Record, 1962). Whilst this conclusion was based primarily on his analysis of the statistics for England and Wales, he also argued that the same arguments could be applied to Sweden, France, Ireland and Hungary (McKeown, Brown, & Record, 1972).

Although McKeown’s work has been hugely influential, it is also highly contentious. Critics have argued that it overestimates the scale of the decline in mortality during the eighteenth century and underestimates the impact of sanitary improvements during the second half of the nineteenth century. Although Simon Szreter (1988, p. 37) argued that ‘McKeown’s work continues to provide the only thoroughly-researched empirical support for the extreme laissez-faire point that health and welfare gains may be generated most effectively merely as a by-product of economic growth’, he also argued that it was profoundly mistaken. He concluded that that the evidence presented by McKeown and his coauthors ‘does not in fact provide … conclusive and extensive support … for the contention that rising living standards and associated nutritional improvements have been the predominant source of mortality decline in Victorian and Edwardian Britain’ (ibid., p. 34).

As Harris (2004, pp. 380–3) pointed out, one of the major issues raised by these debates concerned the relationship between nutrition and infection. McKeown argued that improvements in nutrition reduced mortality because they increased resistance to infection, and he only allocated a secondary role to the impact of sanitary measures which reduced disease exposure. However, as Scrimshaw and SanGiovanni (1997) explained, the relationship is actually more complex. This is because, although improvements in nutrition may lead to increased resistance, reductions in exposure can also lead to improvements in nutritional status. As Floud, Wachter and Gregory (1990, p. 245) explained in their analysis of the history of human height in Britain between 1750 and 1980, children who grow up in disease-ridden environments are likely to retain fewer nutrients. Outbreaks of infectious disease reduce appetite, and the body’s need to fight infection diverts nutritional resources from physical growth.

During recent decades, economists, historians and demographers have approached these issues from a number of different angles. As Szreter (1988, pp. 10–11) argued, one of the limitations of McKeown’s approach was the way in which he reached his conclusions by a process of elimination, whilst providing relatively little evidence that nutritional standards actually improved. During the 1970s and 1980s, Fogel et al. attempted to address this issue with the aid of height data. They argued that the average height of a population reflected its ‘net nutritional status’ and that this provided a source of information about nutritional standards which McKeown and his coauthors lacked (see e.g. Fogel et al., 1978). However, as Fogel recognised, the concept of net nutrition is actually quite complex, and reflects the impact of both diet and disease (see Fogel, 1986, pp. 457-8).

A number of authors, including Fogel, have also tried to address the issue by providing direct estimates of nutritional intake, but these have also proved controversial. Floud et al. (2011, pp. 151–64) argued that the average number of calories available for human consumption in England and Wales rose slowly during the course of the eighteenth century, stagnated between circa 1800 and 1850, and then rose much more rapidly. At the same time, however, other authors have argued either that there was little change in food availability before the 1860s (Broadberry, Campbell, Klein, Overton, & Van Leeuwen, 2015), or that food availability declined dramatically between circa 1770 and 1850 (Meredith & Oxley, 2014). These disagreements reflect differences in the assumptions made by different authors regarding such issues as the amount of land under cultivation, seed ratios, and the proportions of different crops which were either fed to animals, used up during the production process or lost as a result of wastage. Although Harris, Floud, and Hong (2015) have continued to argue that changes in food consumption are likely to have played some part in mortality change, they have also reiterated their view that public health improvements also played a critical role.

Whilst some authors have examined changes in diet and nutrition, others have focused more attention on improvements in public health and hygiene. Peter Razzell (1993, p. 769) argued that the late-seventeenth and early-eighteenth centuries saw major improvements in standards of English housebuilding, leading to ‘a very significant improvement in domestic hygiene’. Alex Mercer (2014, pp. 13–14) and Graham Mooney (2015, p. 153) have highlighted the importance of improvements in domestic and personal hygiene during the middle years of the nineteenth century. Jaadla and Puur (2016) have argued that low-cost improvements in the provision of water supplies had a significant impact on infant mortality in the Estonian University town of Tartu prior to the introduction of either waterworks or sewerage. These arguments resonate with James Riley’s (2008, p. 165) conclusion that the major causes of the decline in mortality in twelve low-income countries during the twentieth century included ‘primary education, inexpensive public health improvement, basic access to rudimentary health care, popular understanding of health risks … and the involvement of people in public health endeavours’.

However, despite this emphasis on low-cost innovations, other authors have focused much greater attention on the major efforts made by public authorities to invest in water supplies and sanitary infrastructure. Cain and Rotella (2001) argued that ‘expenditures on sanitation had a large impact on reducing the waterborne disease death rate’ in 48 United States cities between 1900 and 1929 and Szreter (1988, p. 25–6) identified a direct link between increases in the value of the loans sanctioned by the Local Government Board in England and Wales after 1870 and the onset of a sustained decline in mortality after that date. More recently, Jonathan Chapman (2017, p. 1, 2019, p. 233) has tried to estimate the level of municipal sanitary intervention by examining the value of municipal debts accumulated in England and Wales between 1867 and 1900. He concluded that the investments associated with these debts helped to explain up to sixty per cent of the decline in urban mortality between 1861 and 1900, and approximately 88 per cent of the decline in mortality between 1861 and 1890.

Although these findings suggest that there was a strong relationship between investment in ‘public works’ and changes in mortality, they suffer from a number of limitations. One particular problem is the failure to distinguish between investments in different kinds of activity. However, these issues have been addressed in other studies, both in Britain and elsewhere. Some authors have focused on the sequencing of investments in water supply and waste removal, whilst others have examined specific improvements in water quality.

As John Hassan (1985) demonstrated, the sequencing of investments in different services may have been especially significant. He argued that many English and Welsh municipalities invested in improved water supplies without making equivalent attempts to improve arrangements for sewage removal, with the result that many residents were now at greater risk. William Hubbard (2000) made a similar point when he compared the sequencing of sanitary investments in the two Norwegian cities of Bergen and Oslo. Both cities made efforts to improve their water supplies during the 1860s but only Bergen made comparable efforts to improve arrangements for the removal of waste products. Hubbard argued that this meant that Bergen experienced a much greater decline in mortality from water- and food-borne diseases before the end of the nineteenth century (p. 345).

The complementary nature of improvements in both water supply and sewerage systems has also been highlighted in recent studies of mortality change in France and Germany. Kesztenbaum and Rosenthal (2017, p. 174) found that the establishment of direct connections to sewers had a ‘large and positive [i.e. beneficial]’ impact on mortality in Paris between 1880 and 1914. Gallardo Albarrán (2018, p. 27) found that improvements in the water supply had a much greater impact on the decline of mortality in Germany when they were accompanied by improvements in waste removal arrangements. The combination of improved water supplies and better sewerage reduced mortality by between 19 and 23 per cent between 1877 and 1913.

Although these authors have emphasised the need to avoid contaminating water supplies, others have focused on efforts to improve water quality by positive means. In a widely-cited paper, Cutler and Miller (2005, p. 13) studied the impact of filtration and chlorination on mortality in thirteen United States cities between 1900 and 1936. They found that the introduction of these ‘clean water technologies’ accounted for approximately 74 per cent of the decline in infant mortality and 62 per cent of the decline in child mortality during this period. Alsan and Goldin (2015) argued that these technologies had a much smaller impact on infant mortality in Massachusetts and Cutler and Miller (2016) have recently conceded that their own estimates may have been excessive. Anderson, Charles and Rees (2018, pp. 21–4) have argued that they should be reduced much further, so it seems reasonable to suppose that the debate will continue.⁴

2. Urbanisation and urban mortality

As we have already seen, debates over the causes of mortality change are closely bound up with questions of urbanisation. It is generally agreed that urbanisation exacerbated existing public health problems and this implies that there was a specific need to address the problems associated with such areas. However, as several of the contributions to this special issue will demonstrate, efforts to investigate the impact of ‘urban’ sanitary reforms have also been complicated by the difficulty of defining ‘urban’ areas.

As De Vries (1984, p. 21) explained, the use of such terms as ‘towns’, ‘cities’ and ‘urbanisation’ is far from straightforward. In early-modern Europe, towns were defined by legal and administrative rather than demographic criteria, and this continued to influence the identification of urban locations throughout the period covered by the contributions to this special issue. As Peltola and Saaritsa (2019, this issue) explain, ‘in … Finland, the concept of “city” merits qualification … many of the administrative-statistical units included in this category were small and essentially non-urban compared to the industrial cores of the western world’. The Swedish authorities only adopted a population-based definition of cities in 1965. As late as 1930, eighty per cent of Swedish cities contained fewer than 15,000 people, and the median-sized city had a population of only 7000 (Helgertz & Önnerfors, 2019, this issue).

These issues were not confined to the Nordic countries. As Harris and Hinde (2019, this issue) explain, they also applied, in slightly different ways, to England and Wales. During the late-nineteenth century, the UK government divided England and Wales into approximately 1000 urban sanitary districts and nearly 600 rural sanitary districts,⁵ but these categories were based on a number of different local government units, including boards of health, municipal corporations, Local Government Act Districts and Poor Law Unions. Although the majority of the urban sanitary districts were relatively populous, some – such as the urban sanitary district of Childwall, on the outskirts of Liverpool – were very small. However, when it came to approving the creation of new urban districts, the Government was committed to satisfying itself ‘that each of the proposed district[s] was really urban in its character, or rapidly becoming so, and that among the inhabitants there was the requisite element for an efficient local governing body’ (ibid.).

The difficulties of investigating the impact of urbanisation on mortality in England and Wales are compounded by the plethora of different local government units associated with different functions, which reflected the ‘ad hoc’ way in which these functions had developed (Webb & Webb, 1908, p. 755). One example of this was the distinction between registration districts and sanitary districts. After 1872, the UK Government placed the burden of responsibility for sanitary administration on different types of sanitary district which were themselves based on a variety of older authorities. However, mortality data were collated by the Registrar-General on the basis of registration districts. These were more-or-less coterminous with Poor Law Unions, but these were often quite separate from the units responsible for sanitary administration (Harris & Hinde, 2019, this issue).

Many writers have tried to circumvent the problems posed by contemporary definitions by classifying areas on the basis of population density. One particularly common ‘cut-off point’ involves the identification of areas whose population density exceeded one person per acre, or approximately 250 people per square kilometre. The British poverty researcher, Charles Booth, used this cut-off point when he attempted to classify English and Welsh Poor Law Unions in the 1890s. He estimated that 175 out of a total of 631 Poor Law Unions had a population density of more than one person per acre. However, there were only 112 Poor Law Unions in which more than fifty per cent of the population actually resided within ‘urban areas’ and, even within the 175 ‘urban’ Poor Law Unions, only 67 per cent of the population was actually ‘urban’ (Booth, 1894, pp. 55–104).

In addition to looking at population densities, other authors have also suggested that it is necessary to take account of the population’s occupational structure. De Vries (1984, p. 11) argued that cities could be distinguished from other forms of settlement by ‘population size, density of settlement, share of non-agricultural occupations and diversity of non-agricultural occupations’. He believed that a large mining settlement should not be regarded as a city, even if it fulfilled the first three criteria, because it was insufficiently diverse. However, one might argue that a degree of occupational diversity is less important when considering the specific ways in which the size and density of the population influence its mortality.

One apparently straightforward way to identify urban areas might be to do so on the basis of population size. De Vries (1984, p. 22) believed that ‘a serviceable definition of urban population in early-modern Europe is the inhabitants of densely-housed settlements of at least 2000 or 3000 population’ and that ‘a threshold … [of] say 3000 … would embrace very nearly all of the functionally-urban population’. However, he also claimed that ‘historical evidence of the populations of the many hundreds of such cities is unavailable … [and] at this level it becomes important to know about the occupational attributes, in order to avoid erroneously including non-urban places’. He therefore chose to adopt a threshold of 10,000 inhabitants. Smith, Bennett and Radicic (2019, pp. 576-9) also employed a threshold of 10,000 inhabitants in their recent study of urbanisation in late-Victorian Britain. Using this figure as a starting-point, they estimated that just over seventy per cent of the population of England and Wales could be classified as urban on the eve of the First World War. However, they also suggested that approximately twenty per cent of the population inhabited ‘transitional’ areas, of whom just over one-third (seven per cent of the total) lived in areas with ‘urban characteristics’.

Although this brief review has highlighted the difficulties associated with the identification of ‘urban’ areas, it also raises an important question regarding the relevance of small population centres to the study of urban mortality. As Peltola and Saaritsa (2019, this issue) have suggested, the literature on urbanisation and mortality has often tended to focus on large urban agglomerations, but this does not necessarily mean that smaller or less-densely populated centres were immune to the challenges posed by urbanisation, or that all of these challenges were the same in areas of different sizes. It is also important to recognise that, even at the end of the nineteenth century, most of the people who lived in towns and cities containing more than, say, 2000 inhabitants are likely to have inhabited towns containing fewer than 100,000 people.⁶ These issues are addressed directly by Helgertz and Önnerfors (2019, this issue). They argue that, once a decision had been taken to introduce a new sewerage system or water supply in a smaller area, it could be implemented more quickly and its benefits experienced by the entire population. They suggest that we can therefore estimate the impact of sanitary initiatives in these areas with much greater precision.

Helgertz and Önnerfors (2019, this issue) also identify a number of different pathways associated with the spread of water- and food-borne diseases in urban areas. These include transmission from water or food to people; transmission from person to person; and transmission from person to person via transmission from person to food or water. They argue that the operation of these pathways is linked to the size and density of population centres. They suggest that a focus on smaller and less-densely populated centres might therefore have implications for our understanding of measures to control the spread of water- and food-borne diseases in the urban environment more generally.⁷

Some of these issues are also raised by Hinde and Harris in their comparative study of cause-specific mortality rates in different parts of England and Wales between 1851 and 1910. They distinguish between registration districts with different population densities and districts in which different proportions of the population lived in settlements containing more than ten thousand inhabitants. Although there was a clear relationship between both population density and the proportions of people living in settlements with more than 10,000 inhabitants and death rates from a variety of causes, their findings also suggest that mortality was declining at similar rates across all types of area. Although this special issue is primarily concerned with the decline of urban mortality, these results suggest that a focus on rural mortality might also be necessary (Hinde & Harris, 2019, this issue).

3. The impact of sanitary intervention

This special issue is especially concerned with the impact of improvements in the sanitary infrastructure on urban mortality. Our particular focus is on improvements in the water supply and the development of effective sewerage systems. These improvements were likely to have exerted their greatest impact on health by reducing exposure to water- and food-borne diseases. However, as both Preston and van de Walle (1978) and Ferrie and Troesken (2008) have emphasised, they could also exercise an indirect effect on mortality from other causes.

Many of the earliest reasons for investing in better water supplies were only indirectly concerned with public health. As Hassan (1985, p. 538) explained, the long-term benefits of improved water supplies included ‘reduced fire-risks, lower industrial costs [and] the enhancement of property values’ as well as ‘a healthier workforce’. In England and Wales, much of the investment which took place during the first half of the nineteenth century was primarily concerned with industrial uses, especially in textile areas. It was only later that the focus of attention began to shift more decisively towards public health and sanitary issues (Wohl, 1983, pp. 61–3, 110–2; see also Peltola & Saaritsa, 2019, this issue; Ogasawara & Inoue, 2016, p. 8).

Hassan (1985, p. 543) also highlighted the fact that the provision of an improved water supply could be a mixed blessing if it was not accompanied by the development of an efficient system of waste removal. Much of the initial impetus behind the development of such systems was provided by the miasmatic theory of disease. This theory held that diseases were spread by the ‘bad air’ given off by decomposing organic matter. Reformers such as the British civil servant, Edwin Chadwick, argued that an efficient sewerage system would remove organic waste from urban environments and thus improve air quality (Flinn, 1965, pp. 62–3).

M.W. Flinn (1965, p. 63) argued that the miasmatic theory of disease was a prime example of an erroneous theory with beneficial practical consequences. However, public health reformers gained a much better understanding of the factors associated with the spread of infectious diseases following John Snow’s investigations into the origins of London’s cholera epidemic in 1854 and the subsequent development of the bacteriological theory of disease (Wohl, 1983, pp. 124–5). This led to a renewed focus on the need, not only to improve the supply of water, but also its quality. As a result, a great deal of effort was devoted to the development of new ways of making water not only more accessible, but also safer to drink.

As Cutler and Miller (2005, p. 5) explained, there were two main processes for improving water quality – filtration and chlorination. There were also two types of filtration. Sand filtration involved pouring water into vats filled with sand, gravel and other porous matter. Particulates were strained out of the water as it passed through the sand and gravel, and bacteria were removed either by the formation of a film on the surface of the sand or by oxidisation. The process of rapid filtration involved the use of pressure to force the water through the sand, although some impurities remained. During the early years of the twentieth century, a growing number of towns and cities sought to remove these impurities by chlorination. Although the benefits of chlorination were first demonstrated towards the end of the nineteenth century, there was also a great deal of consumer resistance, and the process only gained widespread acceptance somewhat later (Hardy, 1984, p. 282; see also Peltola & Saaritsa, 2019, this issue).

Although most of the papers in this special issue are concerned with the impact of these initiatives on mortality, they also highlight some of the challenges posed by the kinds of information which are available. These challenges include but are not limited to uncertainties around the dates when particular measures were implemented, the numbers of people affected by them, and the quality of the changes they introduced.

As several of the papers demonstrate, the initiation and implementation of sanitary reforms was often associated with a relatively drawn-out political process. In Sweden, it appears that most reforms were implemented quite quickly, once an initial decision had been reached (Helgertz & Önnerfors, this issue), but this was not necessarily true elsewhere. Hinde and Harris provide an example of this in their discussion of the introduction of a new water supply in the small market town of Sandwich, in southern England. Although the council agreed to appoint a Medical Officer of Health in 1878 and received annual reports on the state of the town’s public health from 1879 onwards, they showed little interest in reforms before the local mayor succumbed to an outbreak of typhoid in 1882. Even then, it took until the end of the decade before they appear to have taken any action (Vaile, 2015). However, in 1891 and 1892, they secured approval from the Local Government Board for a series of loans which were designed to support the construction of a new waterworks, and the new facilities were opened finally in 1894 (Hinde & Harris, 2019, this issue).

This case highlights the value of being able to recreate the political process which generated a particular sanitary initiative, but it is still a relatively isolated one. In their second paper, Harris and Hinde (2019, this issue) describe the construction of a dataset which shows the dates on which either a central government department or Parliament agreed to allow a local authority (or sometimes a private company) to borrow money for public works. However, whilst these data provide a general indication of the chronology of sanitary investment, they do not always provide a specific description of what a particular loan was intended for, and there were sometimes delays between the approval of a loan, the securing of funds, and their practical application. Even in the case of Sandwich, there was a two-year gap between the approval of the final set of loans and the opening of the waterworks for which they were intended.

Some of these issues are addressed by De Looper, Booth, and Baffour (2019) in their study of the impact of sanitary reforms in the Australian city of Sydney. By concentrating on a specific location, they are able to take advantage of detailed information on the installation of a new sewer system and the percentage of households connected to the sewers from the late-1880s. They are also able to identify the specific point in time when a new water supply was introduced. However, they still lack direct information about water quality and the extension of the sewer system to individual suburbs. They also note that the available mortality data are confined to a relatively short time-period.

The identification of independent variables is also discussed by Floris and Staub, Peltola and Saaritsa, and Helgertz and Önnerfors in their studies of Switzerland, Finland and Sweden respectively. Floris and Staub (2019) base their calculations on the years in which a central water supply was first introduced, the renewal of the sewerage system and the interaction between the two. However, they also provide descriptive information showing the years in which the supply of piped water was extended to new districts, and they present additional data on the introduction of bacteriological testing.

Peltola and Saaritsa and Helgertz and Önnerfors also focus on the years in which different reforms were first introduced. Peltola and Saaritsa (2019, this issue) distinguish between the initial supply of piped water, the introduction of sewerage and the use of chlorination to purify the water supply. However, they also emphasise that ‘the initiation of urban sanitation did not mean an overnight regime change, and the modern technologies only gradually pushed out or replaced wells and outhouses’. By contrast, Helgertz and Önnerfors (2019, this issue) suggest that, once reforms had been introduced in Swedish cities, they spread very rapidly to the whole population. Their data suggest that, once a new water supply had been introduced, more than half of all households were connected within a year and more than three-quarters within five years. More than half of all households had also been connected to new sewers within a year of implementation and almost three-quarters had been connected within five years.

The decision to construct water and sewerage systems may have resulted in behavioural changes, whose potential influence on mortality also needs to be considered. For example, if water and sewerage initiatives made towns more attractive places of residence, they may have had a positive effect on migratory flows. If the in-migrants were drawn from a relatively healthy population, mortality might have fallen regardless of the direct effects of any improvements in water or sewerage. Helgertz and Önnerfors (2019, this issue) examine whether migration rates were affected by cities and towns obtaining such services but fail to find any evidence suggesting that this was the case.

There have also been important debates over the choice of dependent variables. As we have already seen, when Cutler and Miller (2005) published their original paper, they paid particular attention to the impact of clean water technologies on infant and child mortality. However, in their response to Anderson et al. (2018), they argued that infant mortality was less likely to respond to improvements in water quality ‘because most infant deaths are birth-related and are concentrated in the first 28 days following birth’ and such deaths were more likely to respond to improvements in obstetric provision than water quality (Cutler & Miller, 2018, p. 9). However, it is important to note that the share of infant deaths occurring in the neonatal period has changed over time. In England and Wales, neonatal deaths only accounted for 34.36 per cent of all infant deaths in the period 1906–10 (Parliamentary Papers, 1919, p. ccxxii).

These issues have important implications for the papers in this special issue. In their study of Finnish mortality, Peltola and Saaritsa (2019, this issue) defend the use of infant mortality as an outcome variable of sanitary investment but the majority of contributors prefer to focus on cause-specific mortality rates. However, the use of such data is complicated by the limitations of contemporary cause-of-death statistics (see also Hardy, 1994) and disagreements over the identification of diseases most likely to be affected.

One of the main problems associated with this question is the limitations of contemporary mortality data. In Switzerland, according to Floris and Staub (2019, this issue), the only available data on mortality from waterborne diseases relate to typhoid, or enteric fever. In England and Wales, typhoid was only distinguished from typhus in 1869, which makes it much more difficult to examine changes in waterborne mortality before and after this date (Harris & Hinde, 2019, this issue; Hinde & Harris, 2019, this issue; Davenport, Satchell, & Shaw-Taylor, 2019, this issue). In Sydney, the terms typhus and typhoid were used interchangeably. However, De Looper, Booth and Baffour (2019, this issue) argue that deaths from typhus were ‘uncommon’ in Australia during this period.

The contributors to this special issue also reflect a variety of views as to the best way to identify the diseases which were most likely to respond to sanitary improvements. As we can see from Table 1, most of the contributors include deaths from either typhus or typhoid (allowing for the problems associated with use of these terms) and cholera. However, Harris and Hinde also include pyrexia, or simple continued fever (Harris & Hinde, 2019, this issue; Hinde & Harris, 2019, this issue), and Helgertz and Önnerfors (2019, this issue) and De Looper, Booth and Baffour (2019, this issue) include gastroenteritis.⁸ Helgertz and Önnerfors (2019, this issue) also include polio, and De Looper, Booth and Baffour (2019, this issue) include deaths from convulsions, teething, atrophy and debility.

Table 1. Dependent variables associated with the impact of sanitary interventions on mortality from water- and food-borne diseases.

Authors	Country	Period	Typhus /typhoid/ enteric fever	Cholera	Diarrhoea & dysentery	Simple continued fever /pyrexia	Polio-myelitis	Gastro-enteritis	Convulsions, teething, atrophy and debility	Infant mortality
Davenport, Satchell and Taylor^a	England and Wales	1832–66		✓
De Looper, Booth and Baffour	Australia (Sydney)	1857–1906	✓	✓	✓			✓	✓
Floris and Staub^b	Switzerland	1876–1901	✓							✓
Harris and Hinde	England and Wales	1817–1914	✓	✓	✓	✓
Helgertz and Önnerfors^c	Sweden	1875–1930	✓	✓	✓		✓	✓
Hinde and Harris	England and Wales	1851–1910	✓	✓	✓	✓
Peltola and Saaritsa	Finland	1870–1938								✓

Notes.

^aDavenport, Satchell and Shaw-Taylor (2019, this issue) also discuss the use of diarrhoea and dysentery but argue that diarrhoeal mortality was not a good index of water quality.

^bFloris and Staub (2019, this issue) argue that improvements in sanitation and water supply should affect mortality from cholera, dysentery and polio, but these headings were not used in their sources.

^cHelgertz and Önnerfors (2019, this issue) also examine the impact of sanitation measures on mortality from a range of airborne infections, as well as its impact on infantmortality and crude mortality, but they argue that improvements in sanitation and water supply were less likely to affect these mortality rates.

The most controversial issue concerns the treatment of deaths from diarrhoea and dysentery, and their relationship to infant mortality. As we can see, most of the contributors argue that deaths from these diseases were directly affected by sanitary conditions, and this is supported by the authors of a recent systematic review, who argue that the incidence of these diseases in the modern world is related to personal hygiene, sanitation and water supply (Cairncross et al., 2010). However, these authors also bemoan the poor quality of much of the evidence for these relationships, and Davenport, Satchell and Shaw-Taylor (2019, this issue) suggest that there was little relationship between the quality of the water supply and the incidence of either infant or diarrhoeal mortality during the second half of the nineteenth century.

In an important paper, Ewald (1991, p. 106) distinguished between water quality and water quantity, and also between a range of infections associated with these issues. He argued that improvements in water quality were most likely to reduce mortality from ‘severe pathogens’ such as ‘classical V. cholera, salmonella typhi and shigella dysenteriae’, and that improvements in water quantity were more likely to reduce mortality from more ‘benign’ pathogens such as ‘el tor V. cholerae, shigella sonnei and enterotoxigenic E. coli’. Davenport et al.’s paper therefore implies that there were significant improvements in the quality of the water supply in England and Wales during the middle years of the nineteenth century. Although they are unable to provide much direct evidence for this, Hassan (1998, p. 23) argued that rudimentary forms of sand filtration were being used to purify the water supply from the 1820s, and Hamlin (1982, p. 59) argued that the Metropolitan Water Act of 1852 led to significant improvements in the quality of London’s water from 1854 onwards.

4. Case studies

All of the papers in this special issue are concerned, directly or indirectly, with the relationship between sanitation and/or water supply and mortality, and all are concerned with the particular problem of urban mortality. However, they approach these issues in a variety of ways. Some contributors focus on the sanitary reforms themselves, whilst others devote more attention to the analysis of mortality trends. De Looper et al. provide a detailed study of one specific location (although they set this within a wider context), but most of the contributors look at a wider range of locations within their chosen countries.

As we have already seen, Cutler and Miller (2005) attached particular weight to the impact of clean water technologies on the quality of the water supply but one might argue that this would only be significant if the untreated water was already contaminated (see also Ogasawara & Inoue, 2016, pp. 20–1). De Looper, Booth and Baffour (2019, this issue) argue that the most important single cause of improvements in the incidence of mortality from waterborne diseases in Sydney was the decision to replace the city’s existing water supply with water from the upper reaches of the Nepean River. The Upper Nepean Scheme was inaugurated in 1886 and, within two years, 85 per cent of the city’s population was connected to it. De Looper et al. argue that this caused an immediate and dramatic step-change in the city’s waterborne mortality.

The comparative importance of improvements in the water supply and the introduction of new sewerage systems is also discussed by Floris and Staub in their examination of the impact of sanitary intervention in seven Swiss cities between 1876 and 1901. They also argue that improvements in the water supply – in the form of central high-pressured clean water – were more important than sewerage but the overall results are rather more modest. They estimate that improvements in the water supply led to a 22 per cent reduction in the typhoid death rate and an eight per cent reduction in infant mortality, but neither set of results was statistically significant and the impact of better sewerage was sometimes negative. Floris and Staub suggest that these results highlight the possible impact of a wide range of additional factors, including other sanitary improvements (including improvements in milk quality), increased attention to personal hygiene, the application of new medical and scientific knowledge, and better diets.

The following two papers assess the effect of sanitary improvements on the decline of mortality in Finland and Sweden respectively. Peltola and Saaritsa examine the impact of improvements in the provision of piped water, the development of new sewerage systems and the advent of chlorination on infant mortality in 37 officially-designated Finnish cities between 1870 and 1938. They argue that each of these developments had a significant effect and that, taken together, they reduced infant mortality by approximately 26 per cent over the course of the period. This meant that they were responsible for approximately 40 per cent of the overall decline in infant mortality within these areas. The authors also suggest that improvements tended to have a greater effect when they were introduced in small- and medium-sized cities towards the end of the period. They suggest that this reflects the importance of context and the extra benefits associated with the introduction of new and improved technologies.

Peltola and Saaritsa also examine the impact of urban sanitary reforms on the history of mortality decline in Finland more generally. They point out that, throughout this period, the majority of the Finnish population continued to reside outside urban areas. This meant that, even though the reforms they describe had a significant impact on the decline of urban mortality, they were less likely to have made a substantial contribution to the decline of mortality in the country as a whole. This suggests that other factors also need to be taken into consideration.

It is particularly interesting to compare these results with those obtained by Helgertz and Önnerfors from their study of Swedish mortality. They measure the impact of the advent of piped water and sewerage together because the two reforms were usually introduced concurrently, and they base their analysis on the experience of 84 ‘official’ cities which installed new water and sewerage systems after 1870. In comparison with Peltola and Saaritsa’s study, they find that sanitary reforms only reduced infant mortality by approximately six per cent between 1875 and 1930. However, they also suggest that these reforms led to a reduction of approximately nine per cent in mortality from waterborne diseases and a five per cent reduction in overall mortality.

As the authors suggest, these results are quite modest when compared with the results reported by other authors. They argue that this may be linked to the relatively small size of the cities in their sample.⁹ However, when Peltola and Saaritsa focused their analysis on ‘small- and medium-sized cities’, they found that the effect was greater (Peltola & Saaritsa, 2019, this issue). This suggests that more attention may need to be paid to the size of urban settlements, the detailed features of their disease environments, and the context and timing of sanitary reforms in future research.

The final three papers are all concerned with the history of sanitary intervention and mortality change in England and Wales. As we have already seen, a number of authors have tried to use the loans contracted by English and Welsh sanitary authorities to measure the extent of sanitary intervention (Chapman, 2019; Millward & Bell, 1998; Szreter, 1988). Harris and Hinde try to take this approach a stage further by providing detailed estimates of the value of the loans which were either sanctioned by central government or approved by Parliament in each year from just after the end of the Napoleonic Wars to the outbreak of the First World War. They argue that the impact of these loans was likely to vary from area to area, depending on local circumstances. However, they also suggest that there is strong prima facie evidence linking the increase in the value of the loans in certain ‘high-performing and high-contributing areas’ during the 1860s to the pace of the decline in waterborne mortality in these areas in the following decades.

In their second paper, Hinde and Harris focus much more closely on the issue of mortality itself. They present new estimates showing the contributions made by different causes of death to the overall increase in life expectancy in different types of area between 1851/60 and 1901/10. Although most of the focus on changes in mortality during this period has been on urban mortality, Hinde and Harris show that there were also substantial improvements in all types of mortality in rural areas. They suggest that, in addition to examining the causes of urban mortality decline, the problem of rural mortality also demands further attention.

As we have already seen, most of the papers seek to measure the impact of sanitary interventions either by looking at infant mortality or at mortality from a range of ‘waterborne diseases’. In their paper, Davenport et al. interrogate this category much more intensively. They suggest that we can gain a deeper insight into the quality of the water supply by focusing on mortality from cholera. This disease became much less important in England and Wales following the second great cholera epidemic of 1848–9, and Davenport et al. argue that this ‘raise[s] the possibility of widespread improvements in water quality’ from the mid-nineteenth century onwards.

Davenport et al.’s paper challenges conventional thinking in two important ways. In the first place, as we have already seen, they argue that the process of improving the quality of the British water supply began rather earlier than has often been assumed. Second, they also argue that the persistence of high rates of infant and diarrhoeal mortality was not directly related to the question of water quality. This suggests that more attention needs to be paid to the aetiology of diarrhoeal mortality and the nature of infant feeding practices (cf. Peltola & Saaritsa, 2019, this issue).

5. Conclusions

As we have already seen, there has been a great deal of interest, in recent years, in the nature and causes of the decline in urban mortality in the late-nineteenth and early-twentieth centuries. Particular attention has been paid to the role played by dietary improvements and, more recently, the control of atmospheric pollution, but the impact of sanitary improvements has also been important. The papers in this special issue have sought to contribute to this endeavour by presenting new evidence relating to the impact of sanitary reform on mortality in England and Wales, Switzerland, Finland, Sweden and Australia.

Although all of the papers are concerned, ultimately, with the relationship between sanitation and mortality, they approach the topic in a variety of ways. Harris and Hinde present new evidence on the value of the loans sought by urban sanitary authorities (and some private organisations) for public works in England and Wales and De Looper et al. describe the inauguration of a new water supply and the expansion of the sewerage network in Sydney, Australia. Floris and Staub, Peltola and Saaritsa, and Helgertz and Önnerfors examine changes in the supply of piped water and the development of sewerage systems in Switzerland, Finland and Sweden respectively. Hinde and Harris present new evidence on the decline of different types of mortality in different parts of England and Wales and Davenport et al. highlight the potential significance of variations in the incidence of mortality from cholera.

The various papers do not provide a single overarching answer to the question of how sanitary reforms affected urban mortality during this period. Harris and Hinde suggest that sanitary investment may have had a significant effect on the decline of mortality in particular British cities and De Looper et al. argue that the introduction of a new water supply had a dramatic effect on mortality from waterborne diseases in Sydney. Peltola and Saaritsa argue that sanitary reforms had significant effects on the decline of infant mortality in urban Finland. However, the impact of sanitary reforms in Sweden and Switzerland appears to have been more limited.

It is difficult to identify any single explanation for these diverse results but the papers do help to identify some of the issues which may be important for future research. Although most of the contributors have attempted to assess the extent of sanitary interventions by using such measures as loans for public works, the development of new water systems or the inauguration of new sewerage networks, there are still significant gaps in our knowledge of the rate at which these initiatives were actually implemented. The papers have also highlighted a lack of consensus over the identification of appropriate outcome variables. Whilst some authors have examined the impact of sanitary reforms on infant mortality, others have focused on specific diseases, and there are continuing disagreements over the relationship between different diseases and different dimensions of the sanitary environment.

This special issue should also help to stimulate further research into the impact of sanitary reforms on towns and cities of different sizes. As Peltola and Saaritsa (2019, this issue) have argued, most of the literature on urban mortality has focused on ‘large and growing cities in industrial countries’. However, they argue that sanitary reforms also had an impact on smaller and less densely-populated areas, and Hinde and Harris highlight the importance of parallel improvements in rural mortality. This suggests that the debate over the causes of mortality decline in all types of area is likely to continue.

Acknowledgement

We should like to thank Angélique Janssens for her help and support throughout this process, and for her comments on this introductory paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. https://data.worldbank.org/indicator/SP.URB.TOTL.IN.ZS (accessed 10 December 2018).

2. The quotations from Plumb and Hicks are taken from Habbakuk (1953).

3. For details of McKeown’s life, see www.oxforddnb.com/view/10.1093/ref:odnb/9780198614128.001.0001/odnb-9780198614128-e-40074 (accessed 8 March 2019).

4. In their response to Anderson et al., Cutler and Miller (2018) acknowledge that their original paper contained a number of ‘unambiguous errors’ which they are happy to correct. However, they also suggest that the main source of disagreement arises from the methods used to estimate the population denominators used to calculate mortality rates. After correcting for the ‘unambiguous errors’ which Anderson et al. identified, they estimate that clean water technologies were responsible for 38 per cent (rather than 41 per cent or 43 per cent) of the total mortality decline. However, they also appear to accept Anderson et al.’s critique of their findings with respect to infant mortality. They now suggest that ‘it is unclear that one would necessarily expect infants to be the demographic subgroup most sensitive to clean water interventions’ (p. 9). In their contribution to this special issue, Davenport, Satchell and Shaw-Taylor (2019, this issue) also propose reasons why the infant mortality rate may have been less affected by issues of water quality.

5. In 1891, there were 1006 urban sanitary districts and 575 rural sanitary districts, but the exact numbers changed over time (Parliamentary Papers, 1893, pp. v-xxxii).

6. The following list shows the numbers of people living in towns and cities containing between 2000 and 100,000 inhabitants as percentages of the numbers of people living in towns and cities containing more than 2000 inhabitants in selected countries in either 1899, 1900 or 1901: Austria (1900): 75.8 per cent; Belgium (1900): 84.6 per cent; Denmark (1901): 57.3 per cent; England and Wales (1901): 43.6 per cent; Finland (1900): 100 per cent; Germany (1900): 70.2 per cent; Italy (1901): 89.4 per cent; Netherlands (1899): 73 per cent; Norway (1900): 62.5 per cent; Sweden (1900): 59.6 per cent; Switzerland (1900): 85.2 per cent. In France, 61.2 per cent of the people who lived in towns and cities with more than 5000 inhabitants in 1901 lived in towns with fewer than 100,000 inhabitants. Sources: (Flora, Kraus, & Pfenning, 1987, pp. 247–81; Bennett, 2012).

7. Ogasawara and Inoue (2016, pp. 18–19) have argued that there is no relationship between the impact of improved water supplies and the size of urban locations. However, their study is based on the analysis of settlements containing more than 20,000 people. As we have already seen, the majority of the areas identified as cities in Finland and Sweden were considerably smaller.

8. Harris and Hinde (2019; see also Hinde & Harris, 2019) excluded deaths associated with this condition because diseases of the stomach and diseases of the digestive system were not listed separately in the Registrar General’s Decennial Supplement in 1901–10. The death rates (deaths per thousand living) for these groups of diseases in previous decades were as follows: 1851–60 (Diseases of the stomach and liver): 1.00; 1861–70 (Diseases of the stomach and liver): 0.98; 1871–80 (Diseases of the digestive system): 0.98; 1881–90 (Diseases of the digestive system): 1.10; 1891–1900 (Diseases of the digestive system): 1.19. The data have been obtained from the Registrar General’s Decennial Supplements for the period 1851/60–1891/1900.

9. They note that their sample excludes many of Sweden’s largest cities, including both Stockholm and Gothenburg, which introduced new water and sewerage systems before the start of their period (Helgertz & Önnerfors, 2019, this issue).