Transport, urbanization and economic development in England and Wales c.1670-1911

Population estimates for England and Wales c.1680

Overview

The Compton census of 1676 and the Hearth Tax returns made in the 1670s and 1680s have been widely used as sources which provide numerical data which can be used to estimate the population at many levels varying from individual parishes to the whole country. Though neither provides data for the whole country, in combination they come close to a full coverage (Figure 1). They therefore offer an opportunity to compare the national population at all levels in the later seventeenth century with the country more than a century later when the first census was taken in 1801. Although both sources have frequently been employed, they have seldom been used conjointly. Attempting to do so has brought to light a range of difficulties which are both frustrating and, perhaps paradoxically, promising.

This is especially the case in relation to the study of urban growth. Over the two-century period from 1600 to 1800 the non-urban population in England rose by 69 per cent; on the continent the comparable figure was 53 per cent, a relatively trivial difference. The urban percentages, in contrast, were 732 and 81 per cent, a massive difference. The far faster rise in the English national population compared to the continent (108 and 55 per cent) was almost entirely due to exceptionally rapid urban growth. Because it is proving difficult and time consuming to reconcile Compton census and Hearth returns for towns, progress has been much slower than we had hoped and anticipated, but the outcome may be a far more authoritative picture of urban totals than currently exist.

Problems with estimating population totals from the Hearth Tax and Compton Census sources

The main problem with using Hearth Tax and Compton Census data to estimate population totals is the uncertainty regarding how to convert the numbers reported in each source to numbers of individuals. The Hearth Tax documents recorded the number of hearths per household, for taxation purposes. The unit reported is therefore the household, and the usual practice is to multiply the number of households reported by a fixed multiplier representing average household size (usually 4.5 or 4.75 persons per household) to give an estimate of inhabitants. However household size was not invariant across the country, and in particular in urban areas the units reported may refer to multiple occupancy dwellings, and so the use of a single multiplier is likely to under-estimate urban populations.

The Compton Census was intended to report the prevalence of religious non-conformity, and the units reported should have referred to the number of adults aged 16 or above. However the instructions to clerics were ambiguous, and it is evident that the units reported could be any of numbers of households, numbers of males aged 16+, adults aged 16+, or all inhabitants. Thus the multiplier to convert these units to population counts could range between 4.5 – 4.75 (in the case of households), 2.86 (assuming ages 16+ comprised 70% of the population, and males half of all adults), 1.43, and 1. Anne Whiteman was able to determine the type of unit reported in a fraction of places, but for most units we have no way of determining the correct multiplier to use.

New estimates of population totals for c.1680

These problems have limited the use of the Hearth Tax and Compton Census in the past. To overcome these problems we have already

1. created a database of all extant Compton Census counts (transcribed from Whiteman, 1986: see Figure 1)
2. created a database of all extant Hearth Tax counts for units not covered by the Compton Census (see Figure 1)
3. spatially matched these units to census units from the 1801 – 1831 censuses
4. compared first-pass estimates of population totals for 1680 with the populations reported for the same units in 1801-31 (Figure 2)
5. compared Hearth Tax and Compton Census returns for Worcestershire and Surrey.
6. created datasets of Hearth Tax counts for Bedfordshire, Buckinghamshire and Leicestershire.
7. compared population estimates from the 1671 Hearth Tax and c.1710 Visitation Returns for Bedfordshire (Figure 3).

The population estimates derived in (4), (5) and (7) should provide a fairly comprehensive coverage of England and Wales, with the exception of the areas identified in Figure 1 as omitted from both sources.

We have already obtained pilot funding from the Cambridge Isaac Newton Trust and the Cambridge Humanities Research Grant scheme to commence objectives 2-6 above. Our estimates of parochial populations from the Hearth Tax and Wake Visitation returns for Bedfordshire parishes (objective 3, above) indicated a very striking level of agreement between the sources, and little population growth in either rural or urban areas in the county over the period 1670-1710 (Figure 3). The Bedfordshire 1671 Hearth Tax Return is unusual in providing counts of paupers as well as tax-payers and those exempted from the tax but not considered pauper. A key objective of future work on the Hearth Tax is to incorporate previously unused sources that enumerate the exempt, something we have already undertaken for Buckinghamshire. For other counties we will spatially match these Hearth Tax data to the units of reporting in our existing database of Compton Census returns, and use the Hearth Tax counts of households to determine the fractions of inhabitants reported in the Compton Census returns. We will then use the estimates of population derived in this way from the Compton Census to create estimates of mean household size for each geographical unit. These estimates will then be used to test whether we can model the relationship between geographical variables and household size with adequate precision and at a sufficiently disaggregated level, for the counties where we already have Hearth Tax data.

We will also use the pilot study to assess the potential to extend this triangulation methodology chronologically and geographically. We have already identified a number of religious censuses with more limited geographical coverage, and have also collected c.7 million baptism and burial records from family history societies, some of which are sufficiently reliable to use for local population estimates.

References

Broad, J. (ed.) (2013) Bishop Wake's summary of visitation returns from the diocese of Lincoln 1706-1715 (2 vols.).

Whiteman, A. (1986) The Compton census of 1676: a critical edition (Oxford, Oxford University Press).
Wrigley, E.A. (2011) The early English censuses (Oxford, Oxford University Press)

Figure 2 shows the change in population between c.1680 and 1801 based on first pass estimates of population c.1680. While the pattern is plausible in many cases, a simple comparison with the population totals estimated previously for the country as a whole and for individual counties (Wrigley, 2011) indicate substantial shortfalls in the population counts derived from the Compton Census and Hearth Tax sources (Table 1). The national total estimated for c.1680 from the Compton Census and Hearth Tax sources was only two-thirds of the national population estimated for 1700 using inverse projection from parish register counts, and was also less than the national population estimate for 1600.

This initial work has made it clear that for many areas we cannot use the Compton Census without comparative data to establish what unit of population was reported. Additionally it appeared (from (5) that the Hearth Tax could produce under-estimates of urban populations if a single household size multiplier were used for all returns. To address these issues we propose

to collect counts of households and hearths from Hearth Tax returns for all regions where they survive and have not been collected by the Hearth Tax Online project; to compare Hearth Tax and Compton Census sources wherever they overlap; to compare these with other sources of population counts available for geographically more restricted areas at similar dates (including Bishop Wake's Summary of Visitation Returns from the Diocese of Lincoln: Broad, 2013) to use a process of 'triangulation' of Hearth Tax, Compton Census and other sources to produce new population estimates for areas where these sources are both available to use where possible baptismal records for the geographical units reported in the Compton Census to identify the fraction of the population reported in the Census and to convert the Census counts to population totals to use the relationship between the population estimates produced in (3) and (4) and the households reported in the Hearth Tax to model the relationship between household size and geographical variables (including urban/rural, elevation, soil type and land-use data); to use the model created in (5) to predict household size in areas covered only by the Hearth Tax, and thus to derive new multipliers to convert household numbers reported in Hearth Tax returns to estimates of population for those areas.

Figure 1. Distribution of surviving Compton Census data, and Hearth Tax data for areas not covered by the Compton Census material.
Sources: Leverhulme Trust project F/09 674/G database of 1670s population returns; Whiteman, 1986.

Figure 2. Percentage change in population between c.1680 and 1801, based on first pass estimates of population in c.1680.
Sources: Leverhulme Trust project F/09 674/G database of 1670s population returns; Whiteman, 1986; 1801 census.
Notes: geographical units of reporting in the Hearth Tax and Compton Census were spatially matched to 1801 census units, and in some cases units were aggregated to ensure a constant unit of comparison.

Figure 3. Comparison of estimates of population for Bedfordshire parishes from the 1671 Hearth Tax and from Bishop Wake's Summary of Visitation Returns. A household multiplier of 4.5 was applied to Hearth Tax counts of households liable or exempt.

Table 1. County population totals estimated by Wrigley (2011) and from Compton Census and Hearth Tax enumerations c.1680.