Methodology

Our aim was to create a new evidential basis for assessing the early phases of the Western epidemiological transition. This transition was defined by Omran as a shift from a phase dominated by ‘pestilence and famine’ to one of ‘receding pandemics’. This shift has long been regarded to have been a combined product of socio-economic, technological and medical influences. However, in an English context limited success has been achieved in providing effective epidemiological explanations for a substantial diminution in the volatility of short-term death rates associated with infectious disease outbreaks between c. 1650 and c.1800. In order to assess the timing and heterogeneity of this phenomenon, we turned to parish burial registers to measure the level of mortality in each month and year in different types of English settlement.

We used abstractions of 4.1 million burials drawn from 621 individual parish burial registers, representing about a 5 % sample of English parishes. This constitutes a substantial enhancement of a dataset of 404 English parishes first developed by Wrigley and Schofield in their pioneering study The Population History of England 1541-1871 (Cambridge University Press, 1981), now extended to incorporate additional parishes from London and other established urban centres as well as newly burgeoning urban-industrial areas.

Map of English parishes from which burials abstractions have been made, showing original 404 parish set in blue and 217 further abstractions in red

English parishes are not uniform in many respects, but especially in terms of their geographical extent and population density. The parishes most appropriate to Wrigley and Schofield’s original objective of establishing the size and growth rate of the English population as a whole were not always those best-suited to considering the volatility of mortality. For example, although the original 404 parish sample is thought to under-represent the smallest rural parishes and hamlets, the small numbers of burials these generate per year makes it difficult to use them in statistically robust comparisons of burial volatility. Much of our efforts in expanding the sample concerned capturing burials from newly industrialising areas such as Leeds, Liverpool and Hull.

The challenge of capturing burials from fast-growing cities

In industrialising cities, explosive population growth fuelled by internal migration was accompanied by rapid expansion of urban centres. The growth of cities strained parochial registration systems by pushing ever-upwards the number of events to be recorded. For burials, there was also the logistical problem of where to put the bodies. Many new churches opened, with associated burial grounds.

In the short term, the building of new churches can produce sharp rises or falls in the burial totals of any one church. These are driven not by changes in mortality but by other churches taking a share of burials previously reported there. When a new church opens, until residents settle into the new arrangements there is a period of adjustment with fluctuations in annual burials that are difficult to disentangle from mortality-driven fluctuations. In new industrial towns the problem is often exacerbated by the fact that the town is not subdivided administratively until a relatively late date, meaning that new churches have no explicit territorial jurisdiction within the urban area and simply take all who come to them.

Since we use fluctuations in burials as a proxy for changes in mortality, we tried to capture burials from all Anglican churches in the urban area in order to aggregate them and smooth out the effect of new churches opening.

Changing share of burials recorded in Liverpool town churches 1662-1812

This entailed the abstraction of large numbers of records. For example, the number of burials in central Liverpool grew from 20-40 per year in the 1660s to 2,500-3,000 per year by 1812, as its population mushroomed from an estimated total of just 1,340 persons in 1670 to some 98,000 persons by 1811, the third largest settlement in Britain. In Leeds and its surrounding suburbs the number of burials grew from 80-120 per year in the 1570s to 1000-1250 per year by 1812. By 1670 its population was already 17,000 and grew to more than 65,000 by 1811.

When does adulthood begin?

To study adult and child burial fluctuations separately, we abstracted 330,000 burials from suitable parish burial registers of those whose parents were mentioned and who were identified as sons or daughters (children) as distinct from other persons without such relationship information (adults). But when a parish register changes from this style of recording to reporting ages numerically, what age should we choose as the end-point of childhood to ensure the best continuity with other data? That is, at what age do sons and daughters stop being identified as such.

Initially we assumed this might be around age 15, but a comparison between the proportion of burials where a parent is mentioned to the proportion of burials where the deceased is said to be aged under 15 from post-1812 Rose Act parish register data showed this to be a little too young, at least for the late eighteenth and early nineteenth century. While there is variation from place to place in this optimum age, it is relatively late, and typically it falls between 16 and 21 years.

This has interesting implications for other purposes besides the analysis of mortality since it is likely to be related to the mean age of leaving home and becoming independent of the parental household.