The Occupational Structure of England and Wales: the 1939 National Register

Social and Economic History

In: Research Data Journal for the Humanities and Social Sciences
Matteo Calabrese Centre for Contemporary and Digital History, University of Luxembourg, Luxembourg, Luxembourg

Search for other papers by Matteo Calabrese in
Current site
Google Scholar
Robin Philips Department of History and Art History, Utrecht University, Utrecht, The Netherlands

Search for other papers by Robin Philips in
Current site
Google Scholar
Robert Keenan Economic and Environmental History Group, Wageningen University, Wageningen, The Netherlands

Search for other papers by Robert Keenan in
Current site
Google Scholar
, and
Bas van Leeuwen International Institute of Social History, Amsterdam, The Netherlands
Utrecht University, Utrecht, The Netherlands

Search for other papers by Bas van Leeuwen in
Current site
Google Scholar
Open Access


A gap in the census surveys for England and Wales between 1921 and 1951 hinders the analysis of their labour structure for the interwar years. The present article uses a dataset containing occupational titles from the National Register – a census-like enumeration of 1939, recently digitised by the genealogy service ‘Find My Past’ – which was previously assigned numerical codes (the pst system). The study expands the existing data analysis on the occupational structure of England and Wales by introducing three further variables: the gender of the surveyed individuals, their age, and the shares of the inactive population per gender and age groups.

Online publication date: 17-10-2022

  1. Related data set “Occupational Structure of England and Wales: 1939 National Register” with handle in repository “Dataverse”

1. Introduction

The occupational structure of England and Wales has received much attention in the past decades. Yet, with the 1931 population census destroyed by a fire, and the planned 1941 census prevented by the Second World War, a data gap exists between the population censuses of 1921 and 1951. The resulting gap has limited our understanding of the English labour structure during the interbellum. A recent contribution filled this void by presenting a first estimate on the occupational structure in 1939 and linking it with the population censuses of 1911 and 1921 on a disaggregated regional level (Philips et al., 2022). In the current article, we build upon this recent reconstruction and extend the data analysis in two respects. First, besides occupation, we further differentiate between gender and age categories – two dimensions which have featured rarely in historical reconstructions of the labour force but which present great value for further research. Second, we assess the shares of the inactive and unemployed population per gender and age categories, based on the 1939 Register enumeration districts.

In Section 2, we introduce the main source used to build the dataset, the National Register of 1939, discuss the issues that this source and its reconstruction present, and finally describe the method used to link the occupational titles listed in the 1939 Register with the primary, secondary, tertiary (pst) classification system. In Section 3, we present our new data, focusing on the subdivisions by gender and age categories, as well as on the weight of the inactive and unemployed. In Section 4, we present our conclusions.

2. The Reconstruction of the Occupation Structure in 1939

2.1. The 1939 National Register

Due to the destruction of the 1931 census and the collapse of plans for a subsequent 1941 census, we lack full regional census data for England during the interwar period. Yet, on 29th September 1939, immediately following the declaration of war against Germany, the British government launched an emergency survey of the population. This survey is known as the “1939 National Register” and includes approximately 42 million entries for England and Wales.1 For each entry, the National Register includes information such as name, year of birth, gender, occupation, and residence by district. Consequently, the National Register data can be considered as comparable to that of a census. However, as the National Register is not formally described as such it does not fall within the usual criteria of publication restrictions of a census – whose publication is allowed only after 100 years. The records of the 1939 Register were subsequently updated during the war (when it was a legal requirement to notify the registration authorities of any change of name or address). But also, after the war, the National Health Service continued updating the Register from 1948 onwards, when it became the basis of the nhs registration system – for which the records were used until 1991 (when records were no longer kept on a paper basis).2

In 2014–2015, the genealogy service Findmypast digitised all the approximately 65,000 transcript books of the National Register in an online dataset.3 Privacy limitations prevent the publication of individual-level data from the National Register. Hence the presented results (and the underlying published dataset) reveal only data on an aggregated level, meaning that they do not include a breakdown from which individual-level information can be extracted. Findmypast created a dataset based on the entered registration forms in the National Register. As such, this article offers added value to the Findmypast dataset by standardising the often inconsistently-entered occupation titles.

Four major issues must be considered for this source. First, the occupational titles are not harmonised, and they depend on the enumerator’s interpretation and transcription of the respondent’s answer – as well on potential errors in the course of copying and indexing.4 The National Register database on Findmypast contains 42,299,296 records associated with 7,040,847 unique occupational titles, therefore presenting us with challenges, often discussed in the literature on historical census analysis, in providing consistent classification and standardisation schemes for a high number of distinct occupation titles (see, e.g., Roberts et al., 2003; Woollard, 2004). Furthermore, around 800,000 of these records contain a blank space for the occupational title and have no birth year, sex or marital status. The Register accounts therefore for c. 41.5 million individuals that have at least one of these indicators.5

Second, a codification analysis has to consider the presence of “undefined” occupational titles, whose sectoral attribution is uncertain. Third, due to the broad existence of areas of informal and unpaid work, the boundaries between the active and inactive population are not always well defined, especially for women performing general service tasks. The same is true when attempting a clear identification of the unemployed population which is often mixed with the inactive population (unless the unemployment status is clearly detectable in the job title).

Finally, the Register deals with the partial exclusion of two occupational groups: 40,000 fishermen not docked at the time of the enumeration (Mitchell, 1988) and 900,000 individuals belonging to armed forces in military installations whose registration was dealt with by the military authorities6 (French, 2000).

2.2. The pst System of Classifying Occupations

In order to solve the above-mentioned issues and create a harmonised dataset, Philips et al. (2022) coded the occupations in the 1939 National Register according to the pst system,7 due to the many advantages of this classification. First, it has already been used to classify the British labour force for the 1379–1911 period (see, e.g., Shaw-Taylor, 2009). Using the pst system therefore facilitates comparisons with earlier occupational data.

Second, through its eight-digit structure, the pst classification allows a twofold approach to the analysis: industry-driven (the first 4 digits of the code – including the empty slots) and occupation-driven (the last 4 digits of the code). By opting for an industry-driven approach in the reconstruction, certain professions, potentially attributable to the tertiary sector, are instead shifted to the primary and the secondary sectors. For example, a worker specifically assigned to manage the accounts in a factory or in an agricultural nursery is coded respectively with the 2,0,0,40 code (“clerk in manufacturing industry”) or the 1,1,1,80 code (“management, farming”). Furthermore, this approach provides an additional tool for solving conflicts of attribution arising from original mistranscriptions of the enumerators or potential errors in the digitisation carried out by Findmypast. For example, while it is unclear whether the titles “Air Ministry Warden” and “Air Ministry Warder” refer to distinct occupations, their industry code is undoubtedly the same, namely 5,50, which indicates “armed forces”.

Third, the pst code has a significant degree of flexibility for dealing with sectorally unspecified workers. For example, some occupational titles contain generic formulas but may be attributable to the tertiary sector – such as “shop owner”, “watchman”, or “laundry services”. Within this group, sub-sectoral attribution can be uncertain with an industry-driven approach. pst allows for this uncertainty with a group of tertiary codes which account for this ambiguity – for example, the 4,0,0,3 code for “shopkeeper”, “shopman” or “shop worker” or the 5,20,4,0 code for “laundry work”. Completely generic occupational titles – those which ascribe occupations in the active population but with an unclear affiliation to one of the three sectors – such as “general labourer” or “heavy worker” can be instead coded using a group of codes starting with the digits “90,0” which identifies them as “sectorally unspecific occupations”.

Finally, the presence of codes for the categorising of the inactive population (starting with the digits “99,3”) allows us to include in the codification those entries with missing or incomplete information in the occupational title – often paired with missing data on age and gender. Entries with missing or partial data on age and gender but with a clearly defined occupational title are coded following the same procedure as for those with complete information – although eventually only considered for the statistics on “total population” and not for the single sub-groups based on gender or age (see, e.g., Section 3.3).

2.3. Two-Step Approach Towards Coding the Occupation Titles

Due to the high number of unique, non-standardised occupation titles, two approaches were adopted to categorise occupation titles. First, the 13,359 occupational titles with a number of observations equal to or greater than 100 were linked to their respective pst code. In this way, 30,989,514 records were coded, that is 73.26 per cent of the overall National Register. The titles of this group include a high number of individuals either without a stated occupation or explicitly out of the labour force (e.g., babies, students or women in “unpaid domestic service”). Furthermore, corrections were made for the active population contained in the National Register and for adjusting the number of fishermen and soldiers based on the above-mentioned literature estimations. At this initial stage, 10,040,071 people as part of the total labour force were counted. Second, a random sample gathered from the remaining unattributed occupation titles was made (coding 20,000 occupational titles into the pst classification system, or approximately 0.2 per cent of the remaining, unattributed 7,027,488 occupation titles), to correct the sample of 13,359 occupational titles.

3. Data

  1. 1939 National Register deposited at Dataverse – Handle:
  2. Temporal coverage: 1939

We report the estimated shares for the occupational structure of England and Wales of Philips et al. (2022) and Mitchell (1988), divided per enumeration years and occupational sub-sectors in Table 1.


The dataset includes the following files, which are used for the analysis below: the absolute number of male employees in 1939 by occupational group at district level; the absolute number of female employees in 1939 by occupational group at district level, also including the Krugman Specialisation Index for all occupational groups; the absolute number of employees in 1939 by occupational group and by age group at national level; the absolute number of employees in 1939 by occupational group and by age group at regional level; all occupational titles used in the 1939 National Register, linked with the most appropriate pst code; and finally a correspondence table with the lists of districts in 1911, 1921 and 1939, and a table with the number of observations missing information on gender, age or specific occupation by region, county and district (for more detailed explanations on these files and tables, please see the “Documentation” file in the dataset).

3.1. Data Subdivided by Gender

In Table 2, we present the reconstructed occupational structure of 1939 by gender categories. Our results confirm the predominance of male labour in the primary sector, in both agricultural labour and activities such as mining, in line with the previous figures found by Mitchell (1988) for the censuses between 1901 and 1931, where the participation of women in the activities in this sector is on average around 0.5%. Male dominance is also the case in the secondary sector, however here the percentage of female participation is statistically relevant as upwards of 22% of female labour was employed in the secondary sector.


The presence of female work is particularly weighted towards textiles, where 14.3% of labour-active women are employed as opposed to 3.9% of men. Female participation is even more striking in the tertiary sector where 69.3% of women are employed in contrast to 37.8% of men.

In Figure 1, we present the shares for female and male employment in textiles and services for all districts.

Figure 1A
Figure 1A

Distribution per district of respectively the male, female and total occupied population in 1939 by textiles, as a share of respectively the male, female and total labour force

Citation: Research Data Journal for the Humanities and Social Sciences 7, 1 (2022) ; 10.1163/24523666-bja10026

Figure 1B
Figure 1B

Distribution per district of respectively the male, female and total occupied population in 1939 by services, as a share of respectively the male, female and total labour force

Citation: Research Data Journal for the Humanities and Social Sciences 7, 1 (2022) ; 10.1163/24523666-bja10026

In the analysis of the labour force, subdivided per gender, we find important regional variations. For example, a large proportion of women are involved in the agricultural sector in districts in western Wales, its eastern border with England and in the area of northeast Cambridgeshire on the eastern coast. This contrasts with otherwise low participation rates in the rest of the country. Although participation rates remain low, female involvement in the mining sector largely reflects the geographic concentration of this labour. Involvement in the secondary sector, in particular textiles, suggests that there is a female textile belt across England which is far denser than its gender-neutral equivalent. Here, female participation remains consistently high from Lancashire in the northwest to Bedfordshire in the East Midlands. Although a similar pattern occurs when we disregard gendered occupations, the density along this belt is less consistent and is more concentrated around centres in Lancashire and Leicestershire.

High volumes of female participation in the tertiary sector are also reflected in our figures, especially those employed in the services sector. Here, there is an almost equal, high density of female labour employment across the whole of the country which can be equated with patterns in male employment in the secondary construction sector.

3.2. Data Subdivided by Age Categories

In Tables 3 and 4, we present the occupational structure for aggregate age categories,8 differentiating between male and female labour force.


For both male and female occupation, we found a higher relative percentage of people employed in agriculture in the groups over 60 years old, compared with those below this threshold. Male occupation groups under the age of 60 are instead more often employed in the secondary and tertiary sectors, typically in “Constructions”, “Services” and “Transport”. Women, on the other hand, have higher rates of employment in the tertiary sector than men for all age groups – around 70% on average, while the rate for men is around 35%. Child labour is also present in our figures. In the age groups “0–4” and “5–9”, we found cases of children employed in all sub-sectors with a higher relative percentage in “Services”. The tertiary sector together with agriculture for males and textiles for females are the main sub-sectors of employment for children in the age group “10–14”.

3.3. Data on the Inactive and/or Unemployed Population

Following the pst system, we categorise the occupational structure in two additional groups: the unemployed and the inactive. Whereas the first group is part of the labour force, for the second group, a particular problem related to the pst categorisation arises, namely that this is a combination of several “other” occupation titles, some of which could be potentially be considered as part of the labour force. So this group includes, for example, people with “No stated occupation” (denoted under the pst 2-digit code 99, 1), or people with “No certain status” (denoted under the pst 2-digit code 99, 3). Whereas the individuals undoubtedly attributable to the category “unemployed” are 345,973 (individuals that have a clear “unemployment” specification in their occupational titles), an estimation of the unemployed within the inactive groups is uncertain.

Taking this limitation into consideration, we show in Table 5 the share of people found in the 1939 National Register coded as “Without occupation or unstated”, divided per gender and aggregate age group.9


In Figure 2, we present per district the share of the population that is inactive and the unemployed.

Figure 2
Figure 2

Distribution per district of the inactive and unemployed population, as a share of respectively the male, female and total population

Citation: Research Data Journal for the Humanities and Social Sciences 7, 1 (2022) ; 10.1163/24523666-bja10026

4. Conclusion

The use of the database presented in this article allows us to fill an existing data gap between the population censuses of 1921 and 1951 – since the 1931 census enumeration schedules were lost in a fire during the Second World War, and the implementation of the census for 1941 was cancelled due to the conflict. Future research should strive towards further exploiting this dataset, which has been used in the current study to analyse the occupational distribution of the population of England and Wales.


The research leading to these results received funding from the European Research Council under the European Union’s Horizon 2020 Programme / erc-StG 637695 – HinDI, as part of the project “The historical dynamics of industrialization in Northwestern Europe and China ca. 1800–2010: a regional interpretation”. The article benefitted from the data collection of the ‘Find My Past’ project (


  • French, D. (2000). Raising Churchill’s army: The British army and the war against Germany 1919–1945. Oxford University Press.

  • Mitchell, B. R. (1988). British historical statistics. Cambridge University Press.

  • Philips, R. C. M., Calabrese, M., Keenan, R., & van Leeuwen, B. (2022). The regional occupational structure in interwar England and Wales. Historical Methods: A Journal of Quantitative and Interdisciplinary History.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Roberts, E., Woollard, M., Ronnander, C., Dillon, L. Y., & Thorvaldsen, G. (2003). Occupational classification in the North Atlantic Population Project. Historical Methods: A Journal of Quantitative and Interdisciplinary History, 36(2), 8996.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Shaw-Taylor, L. (2009). The occupational structure of England and Wales, c.1750–1911. Paper prepared for the inchos workshop held in Cambridge, July 29th-31st 2009 [Working paper].

    • Search Google Scholar
    • Export Citation
  • Woollard, M. (2004). The classification of multiple occupation titles in the 1881 census of England and Wales. Local Population Studies, 72, 3450.

    • Search Google Scholar
    • Export Citation

Scotland, Northern Ireland, the Isle of Man and the Channel Islands have been excluded. The National Register also affected these areas, but their conduction was decentralised and the records are yet to be digitised as they have been for England and Wales.


Findmypast gives an estimation of the population with the “names, addresses, marital statuses, occupations and more” of “over 41,000,000 people” (see This estimation, though, is higher than that presented in the statistical report on the survey “National Register. United Kingdom and Isle of Man. Statistics of Population on 29th September 1939. By Sex, Age and Marital Condition. Report and Tables (London, hmso, 1944). This volume gives an enumerated population for England and Wales as 40,651,706. (p.ix). This discrepancy between the two estimations is most likely attributable to later additions to the Register; to exceptions and anomalies in the counting of the armed forces; and finally to the presence of undisclosed entries (for a detailed discussion of these points, see Precise identification of the excluded groups is however hindered by limited access to individual-level data in the Findmypast dataset.

6 Regarding the estimated 1 million fishermen and military forces excluded from the Register’s enumeration see also “National Register. United Kingdom and Isle of Man. Statistics of Population on 29th September 1939. By Sex, Age and Marital Condition. Report and Tables, London, hmso, 1944, p.ix”.


pst stands for Primary, Secondary, Tertiary. For more information: For more information on the system and any supporting material, we can refer to:


Data on male and female employment, broken down into five-year age groups, can be found in our related master dataset.


Data on the inactive population broken down into five-year age groups can be found in our related master dataset.

Content Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 595 362 36
PDF Views & Downloads 801 563 66