Integrated Census Microdata (I-CeM)

• Home
• Usage
• Access
• Data dictionary
• Census history
• Metadata
• Bibliography
• Acknowledgements

CONPARID - its construction, pitfalls, and also its benefits

Alexander Wakelam (2024)

The revised CONPAR variable is one of the most useful aspects of the new iteration of I-CeM but also one which presents some of the most significant analytical issues. Inherently it is an imposition of the I-CeM curators following the lead of the authors of the censuses but one which is arguably separate from the lived experience of those enumerated. CONPAR serves as a measurement of place over time, but place measured in a way that conforms to neither topographical, historical, cultural, social, or even necessarily human understandings of how an area of land might be thought of as a "place". However, it is a useful tool which, theoretically, can be of use to historians working on Britain beyond the modern census era.

© Copyright L Dyer and licensed for reuse under Creative Commons

What is CONPAR?

The core aim of CONPAR is to provide users with a means of assessing person-led data in a geographically consistent area across the whole historical census period contained in I-CeM. CONPAR manages the impact of the doubling of the population, extensive urban expansion, street improvement or infrastructure expansion, and even land reclamation which fundamentally altered the administrative geography of Britain – still largely rooted in medieval realities. Sleepy villages in 1851 became bustling suburbs or minor towns by 1921; men from the ministry homogenized medieval cities riddled with ancient parishes; hungry urbanities gnawed away at the boundaries of other parishes and occasionally entirely subsumed their neighbours. While many rural parishes were enumerated in 1921 exactly as they had been in 1851, particularly in England it is simply impossible to overlay the administrative geography of one census with its predecessor, let alone merge 1921 with 1851.

Warwickshire 1851 parish boundaries by number included in CONPAR

CONPAR is a history of administrative geography but it is a tool of analysis and cannot be analysed in itself. It might be tempting, for example, to use parish amalgamation as a simple measure of population density or urbanisation. Indeed, a cursory examination of a map of England and Wales in 1851 detailing the number of parishes within each CONPAR unit does flag to the eye towns and cities, particularly those that emerged in the second half of the century. In this map of Warwickshire it is easy to spot Birmingham (the city leaching into neighbouring counties) and Coventry to its east; those with knowledge of the West Midlands may even quickly see smaller towns like Leamington Spa to the south of Coventry. Similarly, the town of Nottingham dominates its county, most of the parishes of the county the sole unit within their respective CONPAR. However, this parochial density is misleading – Nottingham's physical expansion was not as extensive as the administrative merging evident here even in 1921. Papplewick, the northernmost parish included within the "Nottingham and Surrounding" CONPAR unit remains, in 2024, a distinct village outside the urban sprawl of the county town. This distinct place though has proved inseparable from the administrative sprawl of its nearby town and it has thus, within CONPAR, been subsumed and its uniqueness dismissed.

Nottinghamshire 1851 parish boundaries by number included in CONPAR

Regions which had complex parochial hierarchies or geographies - particularly in the north west of England where large parishes contained numerous smaller townships - inevitably suffer from the most blatant simplification. The below map of the number of 1851 parishes in each CONPAR unit (using the new 1921 GIS boundaries to further highlight change) in the Greater Manchester area (note the different scale) exemplifies the problem. A small number of CONPAR units - "Manchester, Salford, and Oldham", "Bury and Surrounding", "Rochdale and Surrounding", "Halifax and Surrounds", "Burnley and Surrounds", and "Colne and Surrounding" - account for 137 1851 and 98 1921 parish units. The populations grouped are profoundly unequal. Additionally, this image highlights an aspect of CONPAR's creation, namely that it was not generated with reference to GIS boundaries but simply based on administrative variation. This occasionally leads to islands of unique, undisturbed parishes, in seas of conglomeration. It is possible users may wish to therefore alter or extend CONPAR to create geographically (rather than purely administratively) consistent units.

Greater Manchester Area 1921 boundaries with number of 1851 parishes within CONPAR units

Genesis and Creation of CONPAR

To understand what CONPAR represents and understand how best to treat it within one's analysis, it is necessary to grasp how it was developed.

The earliest version of the English and Welsh CONPAR (present in I-CeM 1.0) derived from the work of the late Sir Tony Wrigley who manually linked common parishes for the 1851-1891 censuses and then separately linked those for 1901 and 1911. These two CONPARs were constructed with an apparently separate logic, preventing a simple merger, and seemingly attempted to simplify administrative geography wherever possible. The most glaring difference between Wrigley's two CONPARs is the latter's amalgamation of London into just three CONPAR units.

Even if such a merger were reliable, it was determined that appending the 1921 parishes to this system was simply not credibly. A series of legislative work since 1911 had redrawn parish boundaries and created new towns out of ancient communities. The men at the ministry who had taken over the operation of the census from 1919 continued this work with gusto, attempting to create reasonable units of enumeration for a modern age out of medieval parochial boundaries. It was therefore determined that it was necessary to create a new CONPAR which reconciled the parish boundaries of all censuses within I-CeM.

Schürer and Wakelam began with Wrigley's extensive work on the two existing CONPARs, first attempting to unpick parishes where possible and subsequently linking the two administrative geographies together, prioritising the more complex 1851-91 system. This involved not only obvious separation in London or the correction of errors where similarly named places were improperly joined, but also extensive work to complexify linkage in regions where new parishes had been created and then destroyed such as following drainage in Lincolnshire. Additionally, it seems likely that Wrigley elected in some instances to join non-contiguous but close parishes together - usually in rural areas implying an attempt to create units of comparable size / population. These have, wherever possible, been disaggregated or enlarged by adding parishes to ensure contiguousness (whichever was more appropriate) but it is highly probable some examples will persist in the data impacting a tiny number of individuals. Similarly, it seems that in the original CONPAR parishes were not always correctly assigned for all years (e.g. PARISH A, B, and C were listed for all years apart from 1881 when PARISH B, C, and D were listed instead) - despite substantial effort it is likely some errors will occasionally persist and, given that each year of data contained c.12,000 unique parishes, some further errors will have been introduced by Wakelam and Schürer. Users are invited to correct obvious misallocations.

These units were subsequently compared to the unique 1921 geography (the altered boundaries of which were thankfully extensively chronicled in the reports produced under the auspices of the administration of Registrar General Vivian) and a new CONPAR system was created. Determining whether a parish could remain a distinct "parochial unit" across the whole censal period was largely dependent on the decisions of contemporary administrators. If, for example, a town expanded beyond its traditional bounds but without overrunning a neighbouring parish, between censuses parish boundaries might be redrawn, a piece of one parish being sent to another, potentially with a piece of a third parish being given to the "losing" parish to compensate. In this example, all three parishes would have to become a consistent unit as it would be impossible, at parish level, to separate distinct geographical areas that were never shared despite these parishes potentially maintaining distinct identities.

Much if not all the contemporary redrawing was highly sensible. Prittlewell was, in 1851 a small fishing village of 2,462 people. By 1911, the railways had brought thousands to the south end of the parish, the population also rising in nearby Leigh from 1,370 to 7,713, from 455 to 3,954 in the parish of Southchurch, and to 58,759 in Prittlewell itself. By 1921, the area had seen a further growth of c.35,000 people. In 1913 the three parishes were combined into the new municipality of Southend-on-Sea. Part of the parish of Eastwood was also hived off to make Southend a more consistent administrative unit though the remainder of Eastwood remained a distinct area. Four parishes had thus become two by 1921; from the perspective of CONPAR, however, this can only be one consistent unit. While "Eastwood" is a distinct parish in each census, given that it is impossible to identify which individuals in 1851 lived in the part of the parish which would be cannibalised by Southend before the 1921 Census, all constituent parts must therefore be grouped under a single heading for all years.

CONPAR benefits

CONPAR is thus a compromise as is evident in the labelling of units including large towns, frequently described as "X and surrounding". Units do not necessarily have comparable population sizes and densities or geographies. The CONPAR unit "Howden and Surrounding Parishes including reclaimed land" is a particularly striking example. The impact of drainage in the second half of the nineteenth-century disrupted parish boundaries significantly with much subsequent altering. As such, despite the CONPAR unit only accounting for c.8,700 people (c.850 individual parishes contained larger populations) it included 29 unique parishes. They are, however, no less incongruous to reality than the continued imposition of ancient parish distinctions which led to 48 separate units associated with the city of York in 1851 for its population of less than 40,000.

CONPAR is thus a powerful tool for assessing how an "area" broadly (but as narrowly as administratively possibly) conceived varied over a long period of time, allowing researchers for example to assess real population growth unrestrained by boundaries that were likely irrelevant even in 1851. As such, when thinking about "places" some users even studying periods before 1851 may find CONPAR units useful, particularly those dissolving 1851-1921 GIS parish boundaries into CONPAR units that can be thought of as descriptive places that reflect how some pre-19th Century descriptions of residence could occur. For example, those reporting addresses as part of eighteenth-century insolvency processes were much more likely to state they lived in "Coventry", "near Liverpool", or give extremely precise detail of their street or court within a city whose precise location has been lost or requires substantial research time to identify. Being able to map on a national level to "Coventry", to "Liverpool and Surrounding", or to LONDON as a unit may allow researchers to increase the size of their research corpus that is "mappable" within GIS.

Users must always take CONPAR with a grain of salt. As repeatedly stressed, the process of creating and recreating units by multiple scholars over more than a decade, the number of parishes in each year, subjective linkage, the process not having been done with reference to GIS mapping, and the likelihood of human error means that errors will persist (some seeming obvious, even inexcusable to the casual user) and even accurate units are not to be analysed in themselves. However, within the scope of the I-CeM project's resources it is hoped a powerful tool of organising data has been produced, representing a significant improvement on previous editions of CONPAR and allowing for the first time real comparison of "space" across the entire census period.