General errors

Edward Higgs

The collection and publication of data from the census-taking and civil registration processes in the nineteenth and early twentieth centuries was a vast administrative undertaking. Thus, in 1901 the census of England and Wales collected nine pieces of information on no fewer than 32,527, 843 individuals – potentially over a quarter of a billion individual data items. In that year information was also collected on 929,807 births, 259,400 marriages, and 551,585 deaths (Sixty-fourth annual report of the Registrar General, vi, xv). Prior to the introduction of machine tabulation in 1911, all these data were collected and laboriously analysed by hand. Untrained householders filled out census schedules, which (until 1911) were copied into enumerator's books by temporary enumerators, from which temporary census clerks in the General Register Offices (GROs) in Edinburgh and London abstracted information by ticking large sheets (Higgs, 2005, 14–19, 154–5). Similarly, the certificates of births, marriages and deaths produced locally in England and Wales from 1837, and in Scotland from 1855, were forwarded to the GROs and analysed by hand. The results from these processes were then published in volumes, such as the Annual report of the Registrar General and the Census reports, also set by hand. This process was bound to lead to various forms of error.

Errors arose out of the failure of householders to give accurate information in the first place. This might simply reflect a misunderstanding as to the information required, especially since some of the instructions provided were complex, and householders often illiterate. Hence, the numbers of people recorded as having medical disabilities in the Census reports is probably grossly inaccurate because there were no clear instructions as to who was to be regarded as 'blind', or an 'imbecile'. Frequently the householders did not know the information required on the census schedule. Many people, for example, did not know their exact ages, and merely made a rough guess. When doing so there appears to have been a tendency for people to round their ages to multiples of ten and five, to 30, 40, 50, 60, etc., and to a lesser extent to 35, 45, 55, etc. This was at the expense of contiguous years, although there is evidence that there was a greater tendency to round ages down rather than up. It has been suggested that this tendency set in after the age of 20 years (Census of England and Wales, 1881. General report, 17; Mills and Schürer, 74–7). Similarly, many medical practitioners seem to have been incapable of giving scientific terms for causes of death on death certificates, and certainly the GRO in London was forced to return the certificates to them for suitable amendment (Higgs, 2004a, 95).

Most of the people involved in data collection appear to have been conscientious men and women but census enumerators inevitably made slips when copying information from the household schedules into their enumerator's books. Enumerators were not particularly well paid, and some were very slapdash and hurried through the work. This may account for the occasional omission of people, or even whole households, from the returns. Moreover, some errors were inevitable given the poor design of the forms used, especially in the case of the censuses. Thus, the enumerators were supposed to copy ages into a dual column in their books – one for males and one for females. On occasion they put the age in the wrong column, and so the number of males and females on the page calculated at the bottom of the columns was incorrect. Census enumerators were supposed to fill out the numbers of houses, households, and individuals on each page in tables at the front of the enumerator's books, but they often seem to have been unable to do so accurately. Fortunately, many of these miscalculations were picked up in the GRO.

But errors could even occur in the central offices in London and Edinburgh. Here the permanent clerks were often overworked, and the temporary clerks employed to tabulate the censuses often of the lowest quality (Higgs, 2004b, 73–5). Despite very careful checking, there were inevitably errors in the abstraction of data, and the calculation of statistics, some of which were picked up and noted in errata slips. But some probably were not, hence the occasional discovery of differences between statistics derived from the manuscript returns (which still survive), and the figures reported in the Census reports. Additional errors may well have crept into the process of printing and publication, which were not picked up during proof reading.

However, such problems are inevitable in any large scale, manual process of this sort. The errors produced are probably quite minor, or cancelled themselves out in the aggregate. But the smaller the geographical scale, the greater the chance that such errors will cause difficulties.

REFERENCES

Census of England and Wales, 1881, Vol. IV. General Report BPP 1883 LXXX.583. [View this document: England and Wales, Vol. IV. General report, 1881]

Edward Higgs, 'The linguistic construction of social and medical categories in the work of the English General Register Office', in S. Szreter, A. Dharmalingam and H. Sholkamy, eds, The qualitative dimension of quantitative demography (Oxford, 2004a), 86–106.

Edward Higgs, Life, death and statistics: civil registration, censuses and the work of the General Register Office, 1837–1952 (Hatfield, 2004b).

Edward Higgs, Making Sense of the Census Revisited. Census records for England and Wales, 1801–1901 – a handbook for historical researchers (London, London, 2005).

K. Schürer and D. R. Mills, 'Population and demography', in Local Communities in the Victorian Census Enumerators' Books, eds Dennis Mills and Kevin Schürer (Oxford, 1996).

Sixty-fourth annual report of the Registrar General (1901), BPP 1902 XVIII (Cd. 1230) [View this document: Sixty-fourth annual report of the registrar-general ]