Note: This is draft version of this paper. It contains no references, though a complete bibliography is appended before the tables. For a more recent version of this paper follow the link to a PDF of the printed version

THE CLASSIFICATION OF OCCUPATIONS IN THE 1881 BRITISH CENSUS

The most laborious, the most costly, and, after all, perhaps the least satisfactory part of the Census, is that which is concerned with the occupations of the people

Summary

This paper will discuss the importance of classifying occupations both to the original collectors of the occupational data contained within the late-nineteenth century censuses and to present-day historians with particular reference to the 1881 censuses of England and Wales and Scotland. It will describe the method by which occupational data was collected and prepared for classification in 1881. The remainder of the paper will report on the initial stages of a Leverhulme funded project, demonstrating that the tabulations of the classified occupations was remarkably consistent (internally at least) notwithstanding the remarks quoted at the head of this paper. The final section will also give provisional analytical results for a single British registration county.

The importance of classification

The population of the United Kingdom less Ireland in 1881 was just over 30 million. The number of unique entries in the occupation column given in a machine-readable version of the census enumerators' books (CEB) for this year is estimated at around 2 million, though many represent slight variations. In any discussion on the use of occupations in historical research there are two problems concerning the grouping of those occupations into meaningful categories. These problems can be divided into two broad groups-standardisation of occupational titles and the classification of occupations. The standardisation of occupations given in the CEBs fall into two main categories, those that lose information and those that retain the full information found in the original. The standardisation of the frequently occurring terms 'Ag. Lab.' and 'Agr. Lab.' to 'Agricultural Labourer' falls under the second head, whereas the standardisation of 'Baker', 'Master Baker' and 'Master Baker employing 3 men' fall into the former category. It could be argued that standardisation which loses information is a form of classification.

The classification of occupational titles is the grouping of discrete occupational titles under a single heading. For example, those giving the occupations 'Farm Servant' and 'Agricultural Labourer'-obviously different occupational titles-might be classified together under the heading of agricultural workers. In this case, in the late-nineteenth century at least, people giving themselves either of these two occupations would be performing a similar economic function, the main difference being that the former would generally be living in the house of their employer whereas the latter would have been living in a distinct property. Similarly the group we have defined as agricultural workers might be classified along with farmers to form an agricultural class. Given that there are an estimated 2 million unique occupational titles given in the CEBs in 1881, it is clear that these titles might be standardised before they can be put to any meaningful use. Given also that a dictionary of occupations produced for the 1921 census contains some 25,000 distinct occupational titles, it follows that classifying occupational titles into groups of similar occupational titles further increases their usability without losing too much of their original meaning. It should be kept in mind that occupational titles, like other nouns, are ultimately only linguistic codes representing our collective understanding of a concept. By extension occupational titles are in themselves classificatory.

A number of different schemes for the classification of historical occupations have been produced in recent decades. These schemes are to some extent mirrored by the different schemes produced as part of the production of the census in the late-nineteenth century. The following section will describe the value and rationale behind the occupational classification schemes which inform the late-nineteenth-century English and Welsh censuses and their value to the present-day historian will be explained.

Contemporary importance

The contemporary importance of the occupational data is not altogether clear. Detailed occupational data was not collected until the census of 1841 by which time the census was under the administration of the General Register Office (GRO). This census saw the introduction of a classificatory scheme-grouping all people (excluding the unoccupied, wives and children) into twelve classes. This scheme was crude, including a labouring class which lumped together miners, charwomen, coachmen and fisherwomen, and while crude, these classifications were not used generally throughout the reports for that year. The more usual form of reporting the data from the 1841 census was in tables of 877 occupations (for England and Wales), listed alphabetically. In 1851, a new scheme was devised, in which occupations were grouped into 17 classes (and one additional for those with no stated occupation). These classes were broken down into 90 sub-classes (including those classes not sub-divided).

This classification scheme and its modification used in 1861 and 1871 was devised by William Farr, Compiler of Abstracts for the GRO and occasional Assistant Commissioner for the Census. The 1851 classification was described by Farr in the 1851 census report. Here he suggested that the classification was based around five main differences in people's work: skill, talent or intelligence; tools, instruments, machinery or structures; materials; processes and products. The greatest emphasis was given to the materials in which people worked as he felt that the other criteria would follow. This emphasis on the materials in which people worked along with Farr's comments on the value of tabulating classified occupations in the 1851, 1861 and 1871 reports clearly indicate that the primary use of the occupational data in this period was to calculate occupation-specific mortality rates. The 1851 report stated that:

It was held to be desirable, not only to take out the numbers of persons of each sex in each occupation, but the number at each quinquennial period of age; for without this information the relative salubrity of the professions, and a great variety of important questions, cannot be determined

without suggesting what the "variety of important questions" were. The classifications used in 1851 were modified for the 1861 census, and in a long appendix to the 1861 report Farr attempted to explain the reasoning behind this altered classification. However, Farr writing in the 1871 general report made his most telling statement as to the purpose of classifying occupations. He implied that on their own these tabulations, in his opinion, were almost worthless; with the addition of age, calculations could be made of specific mortality, the age structure of different trades, professions of manufacture and age-entry levels for the same. The primary aim, therefore for classifying the occupational data collected in the 1881 census was for actuarial purposes. Medial considerations were almost certainly of secondary importance and it was not until the 1880s that a particular interest in the economic structure was taken, moved forward with the work of Professor Alfred Marshall and Charles Booth.

The classification for the 1881 census was markedly different from its predecessor. Higgs argues that this was because Farr's theories on the spread of disease were becoming acceptable and that there was a growing interest in economic and social research. The classification in use in 1881 certainly did not satisfy all those who attempted to use it for these purposes.

This is not the place to describe the problems inherent within the occupational classifications used within the British censuses, but the comments of the Irish Registrar-General in 1871 (which also apply to 1881) make it clear that there were detractors for the British classification scheme. The main qualm given was about the professional class which included private soldiers, postmen and pew openers-"This classification is plainly repugnant to the common meaning of words and the common intelligence of the country." If the classification is flawed then it should be possible to disaggregate occupations within it. In the case of the British classification scheme this is usually possible.

The purpose for the classification of occupations in the census was thus not merely a technique for making indigestible data more digestible but a means to a further end. Apart from publishing the numbers of people in each county in each occupational class (or order), little use was made of the occupational data collected in the census. The occupational data was, however, used to more effect in the Supplements to the Annual Reports of the Registrar General. However, it was not until after the 1881 census that any comprehensive effort was made to produce statistics of the occupational mortality using anything approaching the classifications put in place at the 1851 census. The Supplement to the twenty-fifth annual report of the Registrar-General published raw data for the numbers of male deaths in each occupational order, sub-order, class and occupation for 1860-1 by age, in England and Wales and the respective sub-divisions, but rates of mortality were only calculated for occupations where there was little uncertainty in the occupational nomenclature as recorded with the registered cause of death. For example, no attempt was made to separate 'Agricultural Labourers' from 'General Labourers'. The following supplement, also authored by Farr, followed the same pattern, but for a selected 62 occupational classes some further detail was given. It was not until the following Supplement that a more detailed examination of occupational mortality was undertaken. Deaths over a three year period (1880-2) were abstracted, "by the same clerks, on the occupation sheets of the same form, as had served for the abstraction of the living in the census of 1881, and the same rules as to the details of the process were observed in each case." However, as with the earlier investigations, owing to a lack of uniformity in the occupational titles recorded in the death registers, some of the occupational groups used in the census were not replicated in this report. From this it can be seen that the initial purpose of the classification of occupations in the nineteenth-century censuses was a device to make each occupation more manageable, within an understandable but potentially arbitrary intellectual context.

It was not until occupational titles were used to categorise the population by social and industrial class that the census reports used the occupational data for any other purpose. This innovation is generally seen solely in terms of intellectual progression, however the introduction of more sophisticated mechanical devices for the tabulation of statistical data may have had some part in this development.

Historical importance

To the economic historian the census is of immense value, especially for the period after 1851. Almost all local and national studies of the economy have used the occupational data provided in the census reports to assist in the description of the economy. This is because the census is the only comprehensive source for data on employment by industrial sector. In fact, because there are no single continuous sources for the other usual economic indicators-income and production, it means, according to Clive Lee, that the occupational data provided in the census, provides "the best single indicator of structural change in the British economy and its component regions". The value of this data to economic historians can not be underestimated, and the efforts that have been made by Charles Booth and Clive Lee (amongst others) to produce time-series data from the original shifting occupational classifications of the census reports are indirect testimony to its value (and indeed, lack of alternatives). There are two problems however, with the aggregations of Booth and Lee (and others). First, it is impossible to fully disaggregate all the published classified totals into totals for unique occupational groups. In 1861, for example, those following the occupation of 'Carter' would have been classified along with others to the class of road carrier. In 1881 those occupied as carters would have been classified either to the classes of road carrier or agricultural labourer; the class would have been allocated depending on whether the individual in question lived in a town or in an agricultural district. This is one example of many hundred slight alterations in the classification schemes between 1861 and 1881; many others can be clearly seen in the Census Office's own attempt to provide a series of figures for the period 1871-91. The second problem is the reliability of the classification of the occupations by the census clerks. Though there is no evidence to suggest that the occupational data in Britain were 'edited', as had recently been suggested for the 1880 U.S. Census of Occupations, there is the potential that the abstraction of occupational data for classification as carried out in Britain in the late-nineteenth century was faulty.

Students of the census will know the value of the published census reports, however, the lack of comparability between the differing occupational classifications makes it difficult to produce accurate comparable occupational totals. On a national level it is unlikely that the margin of error is particularly large and on a smaller geographical scale this margin should not be much greater. But, it must always be remembered that the smallest geographical unit for which occupational class totals were published was the urban sanitary district, and even then only when the population of that district was over 50,000. The CEBs provide our only opportunity to accurately disaggregate the national totals given in the published reports to an individual occupational level and they provide the only opportunity to calculate occupational class totals, or individual occupational totals for smaller geographic units.

The occupational data in the CEBs therefore allow us the possibility of grouping either by industry or occupation, respectively giving us the potential to view the economic or social landscape of Britain. The problem of devising a meaningful classification scheme will not be discussed here, as this is a subject which has been well covered by others. Similarly other problems which face users of manuscript census returns like under- and over-enumeration will not be covered in this paper as they too have been tackled elsewhere, though almost exclusively for the United States. Misreporting of data is also problematic. Some discussion of this problem will be found below but further analysis within this area is needed.

Occupational titles and their use

Before discussing the collection of census data, it is necessary to discuss a number of the problems inherent in the collection of occupational titles, be it for the census or for any other census-type listing. The first, and most important point to make is that, in the case of the census, the occupation given, must only be considered as that in which the individual was principally engaged on the day on which the census was taken, rather than that generally carried out throughout the year. This was understood by the census takers, but is not always considered by more recent users of the census. Also, as with much statistical material, users are often unwilling to examine and understand the provenance and potential imperfections of the raw data provided.

The second issue that must be considered revolves around this problem of imperfection within the raw data. The actual meaning of occupational titles, as given in the CEBs should be considered. Both the descriptions of the occupations and their classification are value-laden. And it is as well to be aware of some of the problems that are inherent within occupational titles themselves. The problems inherent in the allocation of occupational title to a class will be discussed below.

First, we should discuss the term occupation as it applies to the census of 1881. The person completing the schedule was asked to give their "rank, profession or occupation". To most reading this the last of these three definitions would probably have been the most important. It could, however, mean one of two things. First, what occupation do you follow and second, what engages you during the day. It is likely that most readers of the schedule would understand it to be the former. They would consider they were being asked to write down the term which was usually used to generically describe what they did for a living. However, there are numerous examples of occupations recorded within the CEBs which refer more to the type of work done than to the simple generic name by which a member of the labour force went under. This is particularly the case for women. The term 'Housewife' is not, strictly speaking an occupation. It is of course a description (albeit highly abbreviated) for what tasks that woman might have carried out during the working day; ('Help at Home', 'Home Work' and 'Household Duties' are other common examples found in CEBs, though the last of these could also be a euphemism for a 'Domestic Servant'.) It is also the case for men (predominantly) with the term 'Labourer'. In certain instances this may be an occupation, but it is often an occupational term with no precision unless qualified by an adjective. Those of 'Independent Means' and suchlike can also be described as having given 'their occupation' rather than 'an occupation'. The numerous people who describe themselves in such terms are not describing their "rank, profession or occupation", but their means of livelihood. However, the majority of information found in the this column in the CEBs can be said to be accepted occupational descriptions, many of which had been current for centuries.

Simple occupational titles are often confusing. They are generic names for a variety of different tasks oriented, usually, but not always, around the same materials. For example, 'Tailor', an occupation usually taken to mean one who makes clothes; however, within the trade the term specifically means one who makes up or sews together the pieces of cloth that a 'Cutter' has shaped.

It is also important to note that the CEBs do not always reflect the schedule completer's own words. The enumerator, as we shall see later, had ample opportunity to standardise the terms given by individuals into terms which the tabulators would understand. It is highly unlikely that many of those people enumerated in the census as agricultural labourers would have defined themselves simply as such-had not the instructions on the reverse of the schedule compelled them-preferring to use a more detailed description of the way in which they were employed around that time of the year. For someone to complete their schedule correctly he or she must have understood the concept of occupational terminology, and it is clear that many did not.

Occupational titles, as recorded in the census, also have geographical or spatial peculiarities. Different terms have different meanings in different places. The usual occupation to fall foul of this problem is 'Clothier'. The instructions to the clerks tabulating the occupations state that:

'Clothiers' in Yorkshire and other cloth manufacturing districts are frequently Cloth Makers; elsewhere they will generally be understood to mean Clothes Dealer.

This begs the question-how would someone who sold clothes in London, but who had previously worked in Yorkshire describe themselves? The occupation 'Wherryman' is similarly problematic. Generally such an occupation is synonymous with 'Drayman', someone who drives a dray-a low flat vehicle for carrying bulky or heavy loads, but when used in areas with considerable expanses of inland waterway it is synonymous with 'Bargeman', either the master or a member of a crew of a sailing barge. A further example is 'Hind' which is best explained from a dictionary definition:

Hind; a farm servant, often living in his employers house; (i) North of England, takes charge of a pair of horses, which he works, feeds and looks after; (ii) (some parts of the North) assists the ploughman, steward (023) or headman (qv); (iii) East Riding of Yorkshire, man in charge of farm where the employer is not resident; (iv) a general utility man.

This form of ambiguity is almost always resolvable with recourse to the original source. However, the dynamic nature of occupational terminology will always impinge on any late-twentieth-century interpretation of a particular occupational title.

Similar to this geographic problem is a temporal problem. Examples of this sort are difficult to find for the nineteenth-century, neologisms being preferred to the creation of homonyms. The meaning of the occupational title "Pilot" would have been clear in the late-nineteenth century, but since the invention of the aeroplane has become ambiguous. Psychological or perceptional differences in occupational titles are also commonly found. These are generally most hazardous when using a number of historical sources to trace the economic function of an individual. Perhaps we underestimate the ability of the nineteenth-century tradesman to understand the function of the document which he is completing. A census, he might think is a government form, thus he will put down the most generic name for his trade, say-'Brewer'; whereas in a commercial directory his occupation is recorded as 'Brickmaker and Porter Merchant'. Other sources, may record further varients. Thus the perception or understanding of the eventual use of the information may have tempered the term the informant used in recording his or her occupation. This can also be the case within the CEBs. Prostitutes, when incarcerated in prison will often be described as such, however, when on the streets or in brothels they will be hidden under such terms as 'Milliner' or 'Seamstress' or even 'Unfortunate'. Similarly it would be unusual to encounter anyone describing themselves as engaged in other criminal activities, but there are those who ironically describe themseleves as 'Retired Poacher'. (The only exception to this is those that were in prison whose crime is sometimes recorded along with their occupation.) There are also ambiguities which are not specifically geographic, as they are occupationally ambiguous, but the meaning can usually be resolved with recourse to the particular geographic areas. An example is 'Drummer'-either someone who plays a drum or a blacksmith's hammerman. Indeed, 'Hammerman' is another ambiguous occupational title. It can refer to someone who worked a hammer for a blacksmith, but it could also refer to a person in a coal mine who hammered out coal once the seam had been exposed.

There are also a number of miscellaneous problems relating to occupations. The most important of these surrounds the practice of giving multiple occupations. For the 1881 census, as before, householders were instructed on their schedules to list their occupations in order of importance. Consider the occupations 'Magistrate and Farmer of 1000 acres' and 'Farmer of 12 acres and Baker employing 30 men', assuming that in each case the individual has specified the occupation they consider to be most important first. In the former case the latter occupation is probably the one which best describes the economic function of the individual, whereas both could be used to distinguish the social status. The second example is more ambiguous. Given that it is unlikely that a farmer with 12 acres would employ 30 men, we assume that it is the baking function which demands classification. Instructions to abstractors for 1881 ruled that the most important occupation should be taken when classifying, with the exception of farmers and auctioneers. Green et al followed the rule that "the occupational description which conveyed more information took precedence". The order in which people listed their "occupations" in CEB can almost certainly tell us something about the way in which they perceived their role in society.

However, amongst those more humbly occupied there is often an element of prestige rather than economic value involved in their multiple occupations which are pose difficulties in untangling, for example, from Buckinghamshire, 'Baptist Minister Quainton Master Coach Builder Emp 1 Man 2 Boys'. Most multiple occupations, however, cause no real problem as the occupations are similar or would be classified in the same general order. (The exceptions seem usually to concern Publicans of any description, Shoemakers, Grocers and Drapers.) As a rule of thumb, for census material at least, it would seem wise to consider the first occupation as the one to classify. Examples of stranger pairs of occupations, again from Buckingham are 'Publican & Pheasant Breeder', 'Shoe Maker & Coal Merchant', 'Grocer & Chairmaker' and 'Draper & Postmaster & Grocery Emp 6 Men'.

The general problem with multiple occupations is in the distinction (or lack of it) between the main source of income or an individual as against the social status of that individual. Little work has been carried out on the incidence of multiple occupations, though the 1851 report considered those farmers who also gave an additional occupation. A more detailed study of multiple occupations would increase of understanding of the perceptions of work and occupations in the nineteenth century.

Two further areas can be identified which cause problems in the identification of economic or occupational status of individuals. The first, are those that are unemployed. Respondents were asked to state their occupation and if unemployed they have should note that. The other is for part-time workers, which particularly affects the work of women and children. In the latter case, there are numerous examples of children described as a scholar along with a recognised occupational term. Which of the two mutually exclusive classes should be these people be classified to?

Ambiguity in occupational terms is not always location-specific. The term gardener, is taken to mean someone who tends a garden. When dealing with an economic classification we must be able to distinguish between those gardeners who are domestic servants and those who work in so-called market gardens. A dairymaid in a farm performs a different task to a dairymaid in a dairy in a town. The terms sailor and mariner are often interchangeable whereas the two words are not strictly synonyms. It is obviously not always possible to discern the exact meaning of an occupational title, but one should always be aware of the potential for ambiguity, even when they seem straightforward.

The discussion above suggests that we must be wary when using occupational terms given in whatever source we find them. Occupational terms are subject to individual interpretation, and in the census, not only by the occupied person themselves, but by the enumerator too. Occupational terms are temporally, spatially and inherently confusing. We must remember that the occupational terms given in the CEBs are at best only a good approximation of the occupations carried out by individuals, and when classified they can be the cause of a number of errors. Drake gives the example of the 233 reported civil engineers in Somerset in 1841, compared with a total of 854 nationally.

Collection of the Census

The method of collecting information from the general public in the form of a decennial census is well-known, but for the purposes of this paper it is necessary to recapitulate the process in some detail. However, this section will only describe the proceedure concerning the usual householders' schedules. A slightly different procedure was carried out for vessels at sea, larger institutions and members of the Royal Navy. Details of these miscellaneous enumerations can be found elsewhere. An "Act of Parliament for taking the Census of England" received Royal Assent on 7 September 1880. The Census itself was taken on Monday 4 April in the following year. 34,711 enumerators were recruited in the period between those dates to distribute schedules (during the week of 28 March 1881) to each household or tenement and to collect those completed schedules on either 4 or 5 April. Enumerators then had six days (i.e. until 11 April) to enter the details recorded on the schedules into their Enumeration Books, "in strict conformity with the rules given therein".

The instructions given to the householders who had been given the schedules were laboriously printed on the reverse of that form. Half of these instructions referred solely to the column on the form relating to the "rank, profession or occupation" of the householder and their family. Despite the exhortation to the public to read these instructions before filling in these forms, it will be clear to all users of the Enumeration Books that these rules were not carried out to the full. Details of these instructions are given in Appendix 1.

Once the householders had completed these schedules, or had had assistance in completing them from the enumerator (or indeed their neighbours), the enumerator collected up all the schedules he had delivered and proceeded to copy them out into his Enumeration Book.

Thus, even at this stage in the Census collection process the CEBs are a recension of the original householders' schedules. It is clear that enumerators made five types of alterations in their 'copying' process. First, minor errors, such as spelling mistakes, were corrected (though they were probably introduced as well). Second, given that enumerators were responsible for the completion of a substantial proportion of schedules, information was certainly altered reflecting their own local knowledge. Thirdly, information which would not have occurred on the householders' schedule, was added by the enumerator. For example at Ayott St. Lawrence in Hertfordshire a comment was added that "A Tramp slept in Mr Pavitts Barn But Left Before the Enumerator Called". Fourthly, it was not beyond the enumerator to falsify information, though this is hardly likely to be significant. The fifth alteration made by the enumerators was a form of standardisation. In the instructions in each Enumeration Book enumerators were told that "such contractions may be used as 'Ag. Lab.' for agricultural labourer, but care must be taken that the contractions used are such as will be readily understood". There is little information on the editing process undertaken by the enumerators in the process of transcribing information from the schedules to the CEBs as almost no schedules remain. We are able to gain some light from a series that exist for some areas of the sub-district of Llandyrnog in Denbighshire for 1851. Comparison of the schedules and CEBs suggested that comprehensive standardisation took place in the occupational field, especially those employed as agricultural labourers or domestic servants. In these cases the occupational detail seems to have been too great, even for those whose schedules were completed by others on the behalf of the householder. On the other hand those schedules completed by the enumerators were said to conform "quite closely with the instructions on the back." These alterations were not likely to critically affect the final census reports, though it would not be extravagant to suggest that in most cases they enhance rather than degrade the quality of information in the CEBs.

Once the enumerator had copied out the schedules and filled in the opening pages of his Enumeration Book, both were dispatched to the one of the 2,175 local registrars (the officer in charge of the Registrar-General's operations in each registration sub-district) along with his memorandum book. After "examination and revision by the registrar [they] were submitted to the superintendent". The registrar was specifically instructed to ensure that the occupation column was filled in correctly. The superintendent registrar approved and countersigned the Enumeration Books and then forwarded them to the Census Office, a temporary office in Craig's Court, Charing Cross, London, where the next checking procedure was carried out. The completed schedules followed a similar route, but by-passed the superintendent. Thus, both the registrars and the superintendents had the opportunity to revise and correct the Enumeration Books though there is little evidence to suggest that they did in any wholesale fashion. Tillott has inferred that one Sheffield registrar carefully corrected the Enumeration Books from the schedules. He cites the entry for a Matthew Parker which gives his occupation as "Humbug". This was altered to give what was presumably on the original schedule-"Angel or Bishop of the Catholic and Apostolic Church, Victoria Street, Sheffield". However, we should note that it is by no means clear, in this case who altered the Enumeration Book. Again, we presume that these alterations enhance rather than corrupt the original data. One registrar gave evidence to a Treasury Committee saying that registrars were unable to completely check the CEBs against the schedules.

The Enumeration Books were subject to further scrutiny at the Census Office-a process which took some 60 clerks around three months. The instructions "respecting the revision of the Enumeration Books" are emphatic in the purpose for which this was carried out:

The Registrar-General expects that the revision of the Enumeration Books will be conducted in so careful a manner as to render the discovery of errors in the future stages of the work of exceedingly rare occurrence.

During this process a number of changes were made in the Enumeration Books, especially under the column containing information pertaining to occupation. These alterations are summarised in Appendix 2. It was during this process that considerable alterations and annotations were made to the Enumeration Books. Though we cannot be certain that some further amendments and alterations were made later, it was probably at this time that information relating directly to the classification of occupations were made. These annotations may have been added during the process of abstraction either as a reminder to the abstractor or as a means for checking decisions, but it seems more probable that they were added at this stage. The problem of identifying the authors of these corrections was as hard at the event as at present. On being presented with an Enumeration Book with the word "indoor" added to the occupation "Farm Servant", William Ogle, the Registrar-General in 1881, was unsure whether it would have been made by a registrar or an abstractor. On balance he thought the former, but it seems much more likely to have been the latter, as the classificatory importance of the addition would have been lost on the registrar. However, though, as has been noted above, the enumerators may have standardised occupations, these annotations often resolve ambiguity.

In 1881, in the county of Warwickshire, the occupation Fender Fitter, when given in the CEBs, was almost always followed by its classificatory code. This alteration could not have been done by the enumerator, the registrar or the superintendent as they did not have copies of the occupational dictionary-it must have been made at the census office by either the abstractor or one of their supervisors. Another example of an alteration that probably occurred at this stage can be found in Warwickshire where the original enumeration book gave the occupation of one George Saville of Aston as "Proffs of Art Naturalist". Another hand has added "Bird Stuffer". (Note that Bird Stuffer is a classification heading.) The effect of these alterations is not clear, for though the schedules were available at the Census Office it is not known whether they were referred to at this stage in the process, thus any 'revisions' were not tempered by local knowledge and inaccuracies may have been introduced. In the machine-readable version of the CEBs for Leicestershire the following four occupational titles occur:

Range Fitter (Others 21/8)
Range Fitter (House Fitter)
Range Fitter (21/8 Others)
Range Fitter (21/8 O)

It should be explained that order 21, sub-order 8, heading 'others', refers to others in Iron and Steel Manufacture. Whereas 'House Fitter', is found in order 11, sub-order 2 (furniture and fittings) under the heading 'House and Shop Fittings. The dictionary of occupations does not contain this occupation under furniture and fittings but under iron and steel manufacturer, but it seems clear that a decision was made, at some stage, to classify at least one individual described as a Range Fitter to a different class.

Once these revisions had been made the Enumeration Books were passed to a variety of clerks for the production of various tables. During this process clerks tabulated the men and women of the country into 414 occupational classifications by quinquennial age groups. To perform this task the abstracting clerks were given abstracting sheets along with printed instructions as to how they should classify the occupations given in the Enumeration Books. This included a classified list of occupations as well as an alphabetical list. The alphabetical list contains some 8,000 occupational descriptions. Thus, with the aid of this and the classified list of occupations the abstracting clerk should have been able to complete the abstracting sheets. The instructions given to the abstracting clerks are summarised in Appendix 3.

Occupational dictionaries

The occupational census is a series of conundrums, the key of which may be, to a certain extent, in that dictionary which the department is said to have

Alphabetical lists were first provided by the Registrar-General for the abstraction of occupations in 1861. Though a similar classification was drawn up for 1851, no list was created for that year. The dictionary of occupations was, in the first instance, not created for the specific purposes of the CO, but for the GRO as a whole. Viscount Palmerston, then Home Secretary wrote to the Registrar-General, George Graham in August 1854 instructing him to prepare returns giving the chief causes of and average age at death for a variety of different professions. Shortly after Graham wrote to the Treasury asking for money to produce a "sort of dictionary classifying alphabetically the exact terms by which each of the numerous workers in trades should be designated". He stated that "without this assistance I cannot make as good use as I could wish of the facts which are in my possession as to the varying mortality in different professions; information which is desired by the public now that attention is much called to the benefits to be derived by the working classes from life insurance and which should be given on the same principle as the numbers living in each profession are recorded in the census."

This suggests that the classifications for the 1851 Census must be highly suspect if classification was left to the (almost) indiscriminate selection of the abstractors. The 1861 dictionary is a highly incomplete document. It was compiled predominantly from trade directories from London and other large towns. Many obscure agricultural and industrial occupations did not find their way into this dictionary. A revision was made for the 1871 census but there seem to be no copies extant.

For the 1881 Census, a completely new dictionary was compiled, with around twice the number of occupations as in the 1861 version. This was compiled using the earlier versions, along with information compiled from leading employers. A pro-forma letter was sent out in January 1881 asking employers for as "complete a list as you can of the sub-divisions of labour in ... ... ...", and of the titles or designations by which the craftsmen in each several sub-division are distinguished".

The dictionary for 1881 was an improvement on its successors but it had two important failings. First it was not completed in time for the Irish census to use it, having the effect that some headings were different, and delaying the tabulation of occupations in Scotland. Second, and more importantly, a number of alterations were made during the census abstracting and tabulating process, leading to differences in the Scottish and the English and Welsh occupational tables. Between the publication of the dictionary and the census reports a number of headings, including 'Student' and 'Farm Servant (Indoor)' were excised. An edited version of the dictionary was sent from the CO in London in September 1882 to the Scottish Registrar General in Edinburgh, but this document only represents a half-way house. It represented neither the original dictionary nor the finished classification. Supplemental guidelines were despatched along with the dictionary which are also of interest as they often augment, but sometimes contradict earlier abstracting rules. The annotated volume is also useful as it provides a demonstration of some of the alterations in classification from the beginning of the abstraction to the end. (However, the final classification as published in the occupational tables in the Census Reports also differs slightly from the altered Instructions.) The additional rules given in the letter are abstracted in Appendix 4.

The dating of this letter is also significant, as along with these instructions an annotated copy of the list of occupations and their classifications was enclosed. This suggests that while the abstraction of the English and Welsh Enumeration Books was well under way, progress in Scotland had been limited. A possible reason for this may have been due to the fact that in 1871 Scotland had for the first time used a different classification scheme to England and Wales. (It should also be noted that it was not until 1911 that it was seen to be important to use the occupations given in the census to help compile the dictionary though this probably happened between 1881 and 1891.)

Returning to the abstracting process, armed with the index of occupations, the abstractors ticked each person to their correct category on the abstracting sheet, produced totals for sub-districts, which were ultimately aggregated to produce the totals for each registration county and published in the Census Reports.

A further problem occurs at this stage, and at all subsequent stages. However detailed the instructions laid down to enumerators, registrars and clerks were, it is always possible that they were interpreted differently by different people performing the same task. This is especially pertinent in the process of abstraction where it seems most likely that the abstracting clerks were responsible for one (or part) of a county. Their inconsistency in interpreting individual occupations could have seriously affected the final published totals for counties. The example above concerning 'Range Fitters' would be appropriate here.

Before ending this discussion of the accuracy of the data within the Enumeration Books, it is well worth examining the role of the clerks in the process. First, the office in which they worked was described by the Superintendent of Statistics William Ogle as:

very low, fearfully unventilated; they were foetid by the time day was over, quite horrible to go into ... we had perpetual complaints of illness ... two of our staff at the present time [1890] declare that their health was permanently ruined by the foul air....

A further practicality must be dealt with here. The abstracting clerks were not paid a salary, but by the number of ticks they made on the abstracting sheets. The instructions say that the abstractor is recommended to learn the classification as:

the more completely he fixes them in his memory, the more rapidly will he be able to do the work of abstraction; and as the remuneration will be in proportion to the amount and accuracy of the work done, larger will be his earnings.

How it was possible to check their accuracy is unknown, though the process may not have changed from 1861, when William Farr wrote that the extraction of data from the Enumeration Books and their classification was not checked.

In the 1861 report Farr said:

To have ensured perfect accuracy in all the details, the whole of the abstracts of occupations should have been repeated in duplicate and compared; but with the force at our disposal, this could not have been achieved in the prescribed time. Considerable pains were taken to correct evident errors of nomenclature; to instruct, assist and check the abstractors in their work; so that, notwithstanding the difficulties in the way, we look with confidence on all the general results of their labours.

And in the same vein, William Ogle in 1889 writing about the 1881 Census:

... I think it desirable that I should make some remarks upon the institution of comparisons between the occupational figures of successive censuses, because I cannot but feel that persons who have no intimate knowledge of the conditions under which the occupational tables are framed are not unnaturally inclined to ascribe to them very much greater accuracy than in my judgment they deserve. I am not here alluding to the differences in the headings introduced at various times, for such changes are so palpable that no one who is justified in using the tables at all is likely to overlook them; but I mean that, even when a heading has remained throughout unaltered, and has always been intended to denote precisely the same group of occupations, a very considerable amount of difference may nevertheless underlie this apparent identity. For inasmuch as the householders' schedules are filled up by the householders themselves, a preponderating proportion of whom are extremely ignorant men and utterly incapable of accuracy, there are always, even in the case of the most simple industries, a large number of returns so carelessly worded that it is very doubtful under what heading they should be placed; and it is very probable that the views taken as to the tabulation of these doubtful cases may have differed at different censuses, taken as these have been at long intervals, and therefore necessarily to a great extent by fresh hands. Again, irrespective of this cause of possible difference, there is another and more important one. Though precise rules for the tabulation are of course laid down by those in charge of the census, the carrying out of these rules, which cannot but be minute and intricate, has in the main to be left to clerks actually employed in the process of abstraction, and requires on their part great care, considerable intelligence, and scrupulous honesty; and it must be confessed that the mode in which the numerous temporary clerks are gathered together and supplied to the census authorities is scarcely such as to ensure these necessary qualifications. Every effort is made to check the abstractors' work, and in the simpler processes of the census such checks can readily be devised and are sufficiently detective, but in the more complicated processes, such as the abstraction of occupations, no sufficient check other than working the whole occupational abstraction in duplicate can be devised, and such a duplicate abstraction is impracticable on the grounds of time and expense. There is always therefore a considerable risk that in the unchecked portion of the abstraction some unscrupulous clerks may have scamped their work, or some careless dullards have made unintentional but serious blunders. When the entire country is taken into consideration, or the occupational heading is one under which a very large number of workers is included, the mistakes and misplacements, which may be made equally well on one side as on the other, may be assumed to right themselves sufficiently; but in small areas, or in occupations with comparatively few persons engaged in them, such counterbalancing cannot be relied upon; and consequently much caution is necessary in the use of the figures.

The checking which Ogle alludes to above is possibly more a check on the abstractor. Abstractors were paid a piece-rate for their work. At the end of a months grind, a sample of their work was reworked by other clerks. If a fixed error rate per thousand was exceeded the abstracting clerk had his salary docked. It is not clear, that if the work had more than a certain rate of errors, that that clerks' work was wholly reworked.

So, from the householders' schedules to the Census Report, the information on occupation was susceptible to alteration and correction from a number of people. These alterations, were in the main beneficial, but they also damaged the integrity of the data. No one, as far as I know, has suggested that the published returns are infallible, but to recognise their imperfections, we must understand the process by which they were created.

Project description

This lengthy introduction brings us to the main purpose of this paper, to report on a project to compare the published occupational totals with newly classified occupations for a single British county.

In late 1996, a project was established at the University of Essex to work on the Genealogical Society of Utah's machine-readable version of the Enumeration Books for England, Wales and Scotland for 1881. One part of this project was to produce a codebook for all the occupations given in the Enumeration Books, estimated at approximately 2 million. The starting point was the county of Hertfordshire, which had been sent to the History Data Service at the University of Essex in advance of the other counties for testing purposes. Armed with a complete machine-readable version for that county along with a predefined occupational classification, known as the Cambridge Code work was able to progress.

Initial work, however, suggested that the Cambridge Code was not stable enough at the outset of the project to use with complete confidence for occupations in 1881. As such it was decided to code all occupational titles, in the first instance, to the classification used in 1881. The benefits of this method were obvious-first a rough check could be made by comparing the published returns with those from the machine-readable version; i.e. to see whether occupations were being correctly coded and second, because the Cambridge code allows one to compare data from different census classifications, it was considered to be comparatively simple to add this further level of coding at a later date.

Classifying occupations to the 1881 Registrar-General's classification is a relatively simple but time-consuming task. However, there are five areas in which classification "errors" have occurred because it was decided to code the occupational titles in absence of other evidence from the CEBs. These five areas are summed up below:

• gender

In agricultural areas female 'Farm Servants' were generally classified by the 1881 tabulators to 'Domestic Servants'. Without information on gender, these will be allocated in the present classification to 'Indoor Farm Servants'.

• relationship to head of household

'Shop Assistants', without any explicit information as to what kind of shop they worked in are classified to 'General Shopkeepers'. However, if these people were described as servants in the relationship to the head of household column in the CEB, in 1881 they would have been classified to the same trade as their master.

Similarly many domestic servant-gardeners, coachmen, grooms, governesses, waitresses and nurses, etc. have by default been classified into non-domestic categories. If information on head of household was examined, many of these cases would be reallocated.

• head of household's occupation

A number of children, gainfully employed are described in such terms as "Helps father in Shop" or "Helps mother", these would be classified as General Shopkeepers and Unoccupied respectively (the latter on the grounds that it is more likely that the mother in question would be a housewife, though this is by no means certain). With information about the occupation of the head of household these people could be classified to a more appropriate category.

• disability

If a person had a disability it was ignored. Therefore a coal miner, for example, would be classified as a coal miner, even if he was also blind, and thus unlikely to be following his occupation.

• geography

The original rules for classifying 'undefined' manufacturing occupations, especially those in the textile industries, were, to some extent influenced by geography. For example, a weaver in an area with a large proportion of the occupied population involved in cotton manufacture would have been classified as in cotton manufacture unless otherwise stated. Working with a single class for each occupational title over the whole country meant that this rule had to be ignored.

The effects of these problems can be seen in the tables that follow. They are generally slight, but care must be taken when interpreting database figures particularly in the areas of domestic servants, agriculturalists and general shopkeepers. The reverse side of this coin is that there will only be some 2 million occupational titles to classify, rather than an estimated 3 million.

It is admitted that this solution is not satisfactory, but with the time available it was deemed better to attempt to code all occupational titles, and warn users of the machine-readable dictionary of the problems inherent within it. All problems, should be of the variety of adding a parameter to the query, rather than altering the actual occupation itself. (Some of the problems about double occupations and unverifiable occupations will remain though.)

Three further problems concerned the manner in which the machine-readable version was created. First, information that had been added in the CEBs in "another hand", i.e. not that of the enumerator was added to the machine-readable transcription in brackets. Information already in brackets was also transcribed in the same way. In some cases this may be helpful but in others it may a hindrance to a clearer understanding of the occupation carried out by an individual. An example is 'Nurse (Domestic Servant) (SMS)', where SMS stands for subsiduary medical service. For the 1881 classification these classes are mutually exclusive. Second, the lack of punctuation in the machine-readable version sometimes poses problems. Third, in the case of multiple occupations, Higgs suggests that abstractors may have underlined that component which was to be classified. This underlining is not replicated in the machine-readable version.

The method used to classify the occupations was simple in practice. A database of all the unique occupations in a county was created first, then a series of around 450 queries was designed to each extract from this database as many of those occupations which should fall into that classification as possible. Ideally all those occupations would automatically be classified to that particular class, however, simple string queries (i.e. all those occupations which include the phrase "Baker") are unreliable when classifying many occupations; for example the occupation "Bricklayer and Baker" should be coded to Bricklayer not Baker. While many thousand occupational strings could be automatically coded, the majority have to be coded 'by hand'. This obviously prolongs the agony, but it was considered necessary, to some extent, to trade efficiency for accuracy. Even in this process there will be errors, but most of these will be corrected either when Cambridge codes are added, or have been corrected when all the occupations for a county are merged with the 'master table' of occupations once a county is complete. One year into the process, some 600,000 unique occupations have been coded representing some 30% of the estimated total. The rate of progress is currently high enough to anticipate completion of the whole of the 2 million occupations during the summer of 1998.

Isle of Man

The aim of this section is to investigate the reliability of the published reports in the classification of occupations. This preliminary study is based on the Isle of Man rather than a mainland county because of the difficulties in mapping the published occupational tables geographically to the data as provided by the Genealogical Society of Utah. The data provided by the GSU has been (approximately) accumulated by ancient county whereas the published reports are given by registration county. This difference will not be a problem in the future, but at present, the data from all the counties have not been received and read from the GSU, which means that a county with no border problems should be relatively straightforward.

The published population of the Isle of Man was 53,558 (including those accidently enumerated twice-54,089), however the database contained some 56,115 records. By removing all records which related to uninhabited houses or houses being built and all records referring to the names of vessels or institutions this was reduced to a total of 54,134 records relating to individuals. The Census Office noted in an addenda that some 540 fishermen had been enumerated twice. Checking the database version found 511 "double-entries"-those which referred to the men while on board ship were excluded. A further 40 records which referred to people who were absent or entered twice (i.e. crossed out in the CEBs) were also excluded from analysis. However, at the end of this process the database still contained 25 more people than the published reports. It is not clear whether this is a cause of faulty tabulation by the Census Office, or an inability to remove unnecessary records from the database. However, 25 records out of 53,558 is a small enough number to make no difference to for this exercise. Table 1 shows the occupational classification for the Isle of Man as given in the 1881 Census Report along with the count for Isle of Man based on the machine-readable version.

[Table 1]

There are some differences between the database and the published reports, but none is particularly striking. And taking into account the problems mentioned above, it indicates that the strategy adopted is broadly comparable with the census enumerating clerks. It would be well to discuss some of the problems which have arisen, in each of the orders. The most striking discrepancy is in the Unoccupied order, which contains 247 less people in the database than in the published report. This will be discussed later.

[Table 2]

Table 2 shows the numbers of males and females classed into each occupational class within the Indoor Service order. There are two important differences between the database and the published returns: the database contains 51 less domestic coachmen than the report and it contains 105 more domestic servants than the report. The missing coachmen have probably been allocated to the cabman class (which has 49 more people in the database than in the report). The shortfall of 105 domestic servants (48 men and 57 women) is harder to explain, though it is likely that these are both caused by the erroneous classification of farm servants. For example, occupations like 'Farm Servant Dom' have been classified to domestic servants, whereas in 1881 they may have been classified to agricultural labourers.

[Table 3]

The database has 143 additional people in agriculture. For the first of these groups, farmers, the main reason the 58 increase is probably the inclusion of occupational titles in this class like 'Local Prim Methodist Preacher Farmer (of 115 Ac.)'. The rule followed was that all multiple occupations which included the term farmer would automatically be allocated to the farmer class. The slightly larger discrepancy in Farmer's Sons and other male relatives is almost certainly caused by the inclusion in the database of very young sons of farmers. These would be able to be ignored when the CEBs were examined, but in the database it is impossible to separate these. There are 18 more Agricultural labourers in the database than in the published report. However, the gender difference (12 less men and 30 more women) suggests that the ambiguity in the occupational titles for Farm Servants has spread both ways. The missing men may have been allocated to the "general labourer" occupational class; while the additional women should perhaps have been allocated to Domestic Indoor Servants.

[Table 4]

The main only significant difference between the published report and the database for those working with animals is in the Fisherman class. Some of this difference may be due to the additional people found in the database, the remainder is uncertain.

[Table 5]

Table 5 shows the differences in numbers in the database and the printed report for those working in food and drink. The main discrepancy here is in those working as milksellers. This is a further problem caused by not using any other information than that in the occupational title. The rules laid down for milk sellers (and associated occupational titles) make this clear:

056 DOMESTIC INDOOR SERVANT. Include Dairymaid but only if not a Farmer's Dairymaid.

100 FARMER, GRAZIER. Includes Dairymen if listed with agricultural servants.

105 FARM SERVANT, INDOOR. Includes "Dairymaids, &c". if living in a Farmer's House. Note, however that this occupational group was not listed in the report. It seems likely that those listed here would have been allocated to Agricultural Labourers (if male) and Domestic Servants (if female)

223 MILKSELLER, DAIRYMAN. Includes "Cowkeeper (in towns), Purveyor of Asses' and Goats Milk. Milk Hawker, Carrier, Contractor."

These rules are not as clear as they might be, and have the added complication that one of the groups (105) was not used in the published report, though it seems certain that these were completely aggregated with those classified as 'Agricultural Labourer'. At present the solution has been to classify all people to do with the production, distribution and sale of milk to milkseller unless they were specifically designated to an agricultural category, i.e. the word "farmer" or "ag. lab" was also within the occupational title.

Table 6 showing the frequency of general and unspecified workers also suffers from slight differences between the published reports and the database. The only easily explicable discrepancy concerns the additional 10 female machinists, who no doubt are sewing machinists. With information on gender these would be correctly allocated to seamstresses.

The final area in which there are discrepancies between the published tables and the database are in the unoccupied and general category, there being 247 less people allocated to this order in the database than in the published report. There are three reasons for this discrepancy. First, the method used by the abstracting clerks would have been to tick each person off that they considered to be in a class, once whole registration districts had been completed, the sum of those occupied was calculated and the difference between the total population and this became the total for Order 24. For the database each and every occupational title has been classified, including those in this order, which will confer greater accuracy on this figure. Second, it seems clear that the tabulating clerks were not always clear about the procedure for those people who were, at the time of the census, not occupied. Legal men, who noted that they were not practising, should have been allocated to the Legal sub-order. However, in some cases it is clear that the tabulators put these people into the unoccupied category. Given that the main reason for producing these occupational tables was to calculate occupational mortality, it would seem foolish not to include those who were temporarily out of work. However, it is also clear that those people who were "retired" or "superannuated" were classed into this section. For the purposes of this exercise, those people who were described as retired were classified with their working counterparts and latterly tagged to denote that they were retired.

Third, it is conceivable that some occupations which could not be coded by the tabulating clerks have been allocated elsewhere in the table. There is also the possibility that other people in this class will also be removed, once further information can be found about their occupational title. There are very few unclassified occupational titles. For the Isle of Man, only three of importance remain: 'Aeide Woman', 'Badge Porter', 'Harrier'. The first is probably a typographical error, the second is probably a dock porter or labourer, whilst the last is probably someone involved in hare hunting. Presently these have not been classified, but as they each only refer to a single person they do not detract from the arguments presented above.

The proceeding section has shown that there is a high degree of correspondence between the newly classified CEBs and the published reports. The main differences are due to a small number of occupational titles where disambiguation is impossible. It has also demonstrated that the coverage of classification is greater in the database than in the reports giving it a slightly greater level of accuracy. However, for the Isle of Man, at least, it is clear that the classification of the occupations was completed successfully both by the abstractors in the 1880s and in the current project. These results are also interesting because they provide evidence against the suggestion by Elias, Halstead and Prandy that recoding studies at this "three digit level" (using 250-300 occupational categories) are in the region of 70-80 per cent".

Conclusion

This paper has shown how the occupational classification scheme for the 1881 census came into being and how it was implemented; it has also described the way in which a present day project is using the same scheme to classify the same occupations and has presented findings suggesting that, with some errors and omissions, the original process of classification was remarkably true to the occupational dictionary. The further classification of the occupations given in the 1881 CEBs to the Cambridge codes will allow comparative work to be carried out using a variety of other classification schemes. What else of the future? There are plans to extend the current project in a number of different directions. The most obvious of these is to reclassify the 1881 returns to provide comparative employment statistics for each census year post-1881 on at least a county level. Given the problems found in the classification of individual level data this will obviously contain a margin of error. A variety of other summary statistics could also be created. Most notably, regarding multiple occupations and non-employment. The latter is perhaps more interesting given that the census reports are usually quiet about this subject. Further possibilities include a national examination of family structure by occupational groups and occupational succession will be able to be analysed.

A number of miscellaneous studies, at at least a county level, could include an examination of the occupations of those with disabilites and the levels of migration within certain occupational categories and a comparison, on an individual level with court reports to see how criminals before conviction described themselves on their schedules.

A further area of research which has not been carried out to any great deatil concerns those individuals who were classified as 'unoccupied' in the published reports. Given that these individuals in 1881 generally represent over 50% of the population their neglect is perhaps unusual.

A final area of research, which will be facilitated by the classification of the occupations in the 1881 census will be comparative studies with the machine-readable 2% sample for 1851. The final aim of the Leverhulme-funded project is to harmonise the classifications between the 1851 sample and the 1881 database. Given that the 1851 2% sample contains only 36,434 occupations, of which at least a half will already had been classified in the 1881 classifications this should not take long.

Appendix 1

Digest of instructions given on the reverse of the householders' schedules.

• Titles of Peers and persons of rank should be inserted along with any other important office they might hold.
• MPs, Magistrates and Aldermen, along with those holding important offices should state their profession or occupation after their official rank.
• Members of the Army and Navy should state their rank and branch of service while officers should also state whether they are retired or on the active list. All pensioners should be designated as such.
• Church of England clergy should describe themselves as "Rector (or whatever) of ...". Other ministers or priests were expected to name their church or chapel. Occasional preachers were expected to give their ordinary occupation along but "may add "Local Methodist Preacher &c. &c."
• Barristers were asked to denote whether or not they were in practice. Clerks in Solicitors' Offices were expected to state what type of clerk they were (Managing, Articled or General).
• Members of the medical profession were asked to state in which branch of medicine they practised (or if they didn't) along with "the University or Society of which they are Graduates, Fellows or Licentiates".
• Professors, Teachers, Public Writers, Authors and Scientific men were asked to state which branch of Science or Art they followed, similarly artists were to "state the art which they follow" and graduates were to give details of their degree.
• Only Medical, Legal or Theology students were to be explicitly returned.
• The term Scholar should only be used to denote a child (or young person) "attending a school, or receiving regular instruction".
• Farmers were asked to state the number of acres occupied, along with the number of men, women and boys employed at the time of the Census. Farmers were also asked to return their children as farmer's sons or daughters. "Men employed on the farm and sleeping in the Farmers's house must be described in the schedule as Farm Servants."
• All others employed on farms but not living in the farmers' house should be described as "Agricultural Labourers, Shepherds &c."
• Merchants, Brokers, Traders, Commercial Clerks and Travellers and Shopmen were asked to state the particular branch of commerce in which they were engaged.
• Master traders and manufacturers were asked to state their occupation in the following style: "Carpenter-Master, employing 6 men and 2 boys".
• Manufacturing trades considered to be "much sub-divided" should return both the branch and the trade, if they were not implied in the name.
• Miners should always describe the mine in which they work.
• Engineers: Civil and Mining Engineers should be explicitly described as such. Factory workmen should be distinctly described: "Engine Smith at Factory". Engine Drivers, Stokers and Firemen should be described in connection with the manufactory, railway &c. with which they were employed.
• Artisans and Mechanics should invariably state their particular branch of mechanical art or business. (original italics)
• The term weaver should not be used without qualification.
• Domestic Servants should be described according to the nature of their service.
• Messengers, porters and labourers should be described according to the nature of their service. The term labourer should never be used without qualification.
• Persons currently unemployed should be described as such, i.e. Coal Miner, unemployed.
• Those following no profession, trade or calling, and holding no public office, but deriving their incomes chiefly from land, houses, dividends, interest of money, annuities &c., may describe themselves accordingly. Those retired should state their former profession. The terms Gentleman and Esquire should not be used.
• Women and children who are employed at home should distinctly record their following.

Appendix 2

Instructions given to clerks at Census Office before abstraction

• Married women returned with a domestic occupation (e.g. Housekeeper), whose husbands had a non-domestic occupation were to have their occupation struck out. (However, terms like Housewife, Cook, Charwoman and Laundress were not to be struck out.) (If the husband also had a domestic occupation, however, this would not have been struck out).
• Farmer's sons (and other near male relatives), aged 15 or above, who had not been enumerated with an occupation were to have the term Farmer's Son added.
• Children under the age of three enumerated as scholars should have their "occupation" struck out.
• All individuals returned without an occupation but containing a description such as "Student, Scholar, Barmaid, Waitress, Cook, Servant &c." in column 6 (i.e. relationship to the head of household) of the Enumeration Book would be given that occupation.
• Men, in business houses (e.g. a Baker or Grocer) would, if they were described as a Servant or an Apprentice in column 6 have an occupation inserted in added to column 10 along the lines of Baker's Assistant or Baker's Apprentice.
• Domestic servants living in farm houses (specifically including "Groom") must be returned as originally stated and not as a Farm Servant. However, the description "in-door" should be added to those people described explicitly as farm servants. Other so-called servants (as described in column 6 (relationship to head of household) living within a farm should have, if their occupation was blank, have their occupation recorded as 'Farm Servant, in-door'.

Appendix 3

Abstrators' instructions.

• Multiple Occupations. "The general rule in such case (sic) is to select for ticking that occupation which would seem the main or more important one. If there be none such, the first entry is to be selected." With the following exceptions - people described as Clergyman and Schoolmaster should always be allocated to the group for schoolmasters. An MP or Magistrate in a profession or branch of industry should be ticked to the latter, and any combinatory occupational title containing the phrase auctioneer should go to auctioneer regardless.
• Men returned by Rank, Degree or Property only should always be allocated to Unoccupied. However, those, property owners who work their property are to be ticked to their specific occupation, e.g. Ship Owner. The instructions state, "in this matter, common sense must be the guide".
• People who were returned by letters, either degrees or membership of a learned society, should be enumerated to the appropriate category.
• Foreign officials are enumerated as Unoccupied.
• Merchants in a specific commodity are ticked "by the nature of the article in which they deal". Foreign merchants are ticked to "Merchants".
• Agents and brokers by default are ticked to that category, with a number of exceptions.
• Farmer's sons or other close male relatives aged over 15 were ticked to class 101 (i.e. 'Farmer's sons'), whereas women with a similar relation were not to be ticked at all
• Cottagers by default were ticked to agricultural labourers (103) though if they occupied and cultivated their land (i.e. had an occupational description such as "Cottager with five acres of land") they should be ticked to farmer.
• Town carters were to be ticked to Carman, but carters in rural districts or small towns were to be ticked Agricultural Labourers.
• All clerks were to be ticked to Commercial Clerks with a number of exceptions (Civil Service, Army, Navy, Law, Shorthand, Banks, Insurance, Railway, Bookstall and Clerks of Works). However, Book-keepers, Cashiers and (non-private) Secretaries were to be considered clerks.
• Apprentices, Journeymen and Assistants were to be ticked to the occupation to which they assist or are apprenticed to. If their occupation was not clear they were ticked to "Indefinite" (22/2).
• Messengers, Errand-Boys and Porters and Watchmen (excepting Railway and Government) were to be ticked to (6/5) and not to their respective trades.
• Warehousemen. First, warehousemen within an industry or manufactory were ticked to that industry. Second, warehousemen when used as a synonym for dealer or shopkeeper should be ticked to the type of goods dealt in. Third, warehousemen, qua store-house keepers were ticked to "Warehousemen".
• "Workmen having any distinct trade or occupation of their own with a Heading for it on the Abstract Sheet, but carrying it on in some manufactory, institution, or house of business, Government or other, are to be ticked to their own occupation or trade, and not to the manufacture to which they are temporarily, or even permanently, attached.
• Labourers working in a distinct industry or manufacture should be ticked to it. Labourers and General Labourers were ticked to 22/2. Special headings were provided for Agricultural, Road and Railway labourers.
• Tenders, Drivers, Stokers and Firemen of Engines (not agricultural, nor on railways, nor in ships) were to be ticked to the heading 22/2, independently of the place or manufactory in which they work.
• For trade designations which occur in distinct industries, e.g. Finisher, Spinner, etc. the instructions suggest that the Enumeration Books should give the "nature of manufacture". Where it was not given the abstractors had three paths to follow, "supply the missing information" themselves, ask the Superintendent, "when ... there is one predominating or exclusive manufacture in the sub-district with which he [the abstractor] is dealing" and thirdly to "tick vaguely designated workmen to the Indefinite groups of workers."
• Retired people were to be ticked to order 24 with the exception of retired Army and Navy officers (my italics), paupers, inmates of workhouses and prisoners-all to be ticked to their former business. Those described in the occupation column as idiots, imbeciles, lunatics or Almsmen/women and Inmates of Almshouses should be considered as retired.
• The unemployed were to be ticked to their usual profession if stated.
• Finally the abstractors were instructed to be on their guard for regional differences, especially where the same name designates different occupations. The example given is 'Clothier'

Appendix 4

Some Additional rules

• "In abstracting occupations it is necessary to view trade as combining many divisions and it is of paramount importance to distinguish the actual work or sub-division of work in which a workman is employed or business in which a shop-keeper is engaged-without reference to-or being biased by-the place in which it is carried on, or the Master, Company or Municipality who pays".
• "The actual Manufacture as represented by the producer (including in many instances the Merchant) whether described as Mechanic, Apprentice, Improver, Assistant, Labourer, Dealer, or Maker."
• "Accessories not so engaged, as Clerks &c. when they do not form the staple labour of the Business (as in Banks, Insurance &c.)"."Distributors of finished Articles, as Carriers, Carters, Carmen, Porters, Messengers, Errand Boys, &c., when they do not form the staple labour (as Railway Porters &c.)"
• "Substantive Trades employed in any business, or trade temporarily or permanently but having a distinct trade of their own (such as Carpenter, or Smith in a Brewery, Hotel, &c."
• "Soldiers and Seamen even if described also as Artizans or Domestics are primarily Soldiers or Seamen, and should be so understood." (italics original).
• "As a rule a Profession, or a Mechanical Trade should be taken in preference to a Dealing or Shopkeeping business if a person be described as both, thus:- John Jones, Blacksmith and Grocer." (original italics) "When Masters or Companies are engaged in several distinct businesses or trades the Master can only be placed to the most important one, but their trade assistants, or trade servants must be placed to any of the several trades in which they may be occupied and domestic servants of the same master would be placed in the domestic class." (original italics)
• "By Civil or Municipal Service is generally meant those persons who are engaged in the general government either of Towns &c. or the Imperial Civil Service; it does not include Labourers, Mechanics, or Scientists or persons of a similar character in professions or trades which may be carried on in Commercial life, as Solicitors, Surgeons, Architects, Telegraphists, Waterworks, or Gas Service, Vestry, Road or Workhouse Labourer, &c., &c., although they may be paid either by the Government or the Municipality".

Bibliography

Official publications

1851 Census of Great Britain, Population tables II. Ages, civil condition, occupations, and birth-places of the people. Vol. 1, BPP 1852-3 LXXXVIII.

1861 Census of England and Wales, Vol. III, General report, BPP 1863 LIII.

1871 Census of England and Wales, General report (Vol. IV), BPP 1873 LXXI.

1881 Census of England and Wales, Vol. III, Ages, Conditions as to marriage; occupations and birthplaces of people, BPP 1883 LXXX.

1881 Census of England and Wales, Vol. IV, General report and tables, BPP 1883 LXXX.

1881 Census of Scotland, Ninth Decennial Census Vol. II, Ages, Education, Civil Condition, Birthplaces, General Index, BPP 1883 LXXXI.

1911 Census of England and Wales, Vol. X, Appendix. Classified and alphabetical list of occupations, BPP 1915, 7660.

Supplement to the twenty-fifth annual report of the registrar-general (London, 1864).

Supplement to the thirty-fifth annual report of the registrar-general (London, 1875).

Supplement to the forty-fifth annual report of the registrar-general (London, 1885).

Occupations of the people (England and Wales) enumerated in 1871, 1881 and 1891, BPP 1895 LXXX [468].

Report of the committee appointed by the treasury in inquire into certain questions connected with the taking of the census, BPP 1890 LVIII.

Other publications

W. A. Armstrong, 'The use of information about occupations', in Nineteenth-century society: essays in the use of quantitative methods for the study of social history, ed., E. A. Wrigley (Cambridge, 1972), 191-310.

R. M. & G. A. Benwell, 'The 1851 census in the Llandyrnog sub-district', Denbighshire Historical Transactions, 27 (1978), 199-201.

R. M. & G. A. Benwell, 'Interpreting the census returns for rural Anglesey and Llyn', Anglesey Antiquarian Society Transactions (1973), 111-36.

R. M. & G. A. Benwell, 'The Llandyrnog householder's schedules for the 1851 census', Local Population Studies, 28 (1982), 89-90.

C. Booth, 'Occupations of the people of the United Kingdom, 1801-1881', Journal of the Royal Statistical Society, 49 (1886), 314-444.

D. J. Butler, Durham city. The 1851 census (Durham, 1992).

M. Campbell-Kelly, 'Information technology and organizational change in the British census, 1801-1911', Information Systems Research, 17 (1996), 22-36.

D. Caradog Jones, 'Some notes on the census of occupations for England and Wales', Journal of the Royal Statistical Society, 78 (1915), 55-81.

S. B. Carter & R. Sutch, 'Fixing the facts: editing of the 1880 U.S. census of occupations with implications for long-term labor-force trends and the sociology of official statistics', Historical Methods, 29 (1996), 5-24.

B. Collins and M. Anderson, 'The administration of the 1851 census in the county of East Lothian', Local Population Studies, 20 (1978), 32-7.

M. A. Conk, 'Accuracy, efficiency and bias: the interpretation of women's work in the U.S. census of occupations, 1890-1940, Historical Methods, 14 (1981), 65-72.

C. A. Crompton, 'Changes in rural service occupations during the nineteenth century. An evaluation of two sources for Hertfordshire, England', Rural History, 6 (1995), 193-203.

G. Crossick, 'From the gentleman to the residuum: languages of social description in Victorian Britain', in Language, history and class ed., P. J. Corfield, (Oxford, 1991), 150-78.

C. Day, 'The distribution of industrial occupations in England, 1841-1861], Transactions of the Connecticut Academy of Arts and Sciences, 28 (1927), 79-235.

M. Drake, 'The census, 1801-1891', in Nineteenth-century society, ed., E. A. Wrigley (Cambridge, 1972), 7-46.

P. Elias, K. Halstead and K. Prandy, Computer assisted standard occupational coding, (London, 1993).

D. A. Gatley, 'Computerising the 1861 Census Abstracts and Vital Registration Statistics', Local Population Studies, 58 (1997), 37-47.

Genealogical Society of Utah, Federation of Family Historians, 1881 Census for Great Britain [computer file]. Colchester, Essex: The Data Archive [distributor], 29 July 1997. SN 3643.

P. Glennie, 'Distinguishing Men's trades'. Occupational sources and debates for pre-census England (1990) (Historical Geography Research Series, 25).

E. M. Green, C. Harvey and P. J. Corfield, The Westminster historical database. Voters, social structure and electoral behaviour (Bristol, 1998, forthcoming).

C. J. Koepp and S. L. Kaplan, 'Introduction', in Work in France. Representations, meaning, organisation and practice, eds, S. L. Kaplan and C. J. Koepp (Cornell UP, Ithica, 1986).

E. Higgs, 'The struggle for the occupational census, 1841-1911', in Government and expertise: specialists, administrators and professionals, 1860-1914, ed., R. M. McLeod (Cambridge, 1988), 73-86.

E. Higgs, 'Structuring the past. The occupational and household classification of the nineteenth-century census data', in History and Computing III, eds, E. Mawdsley et al (Manchester, 1990), 67-73.

E. Higgs, A clearer sense of the census. The Victorian censuses and historical research (London, 1996).

E. Higgs, 'Occupational censuses and the agricultural workforce in Victorian England and Wales, Economic History Review, 48 (1995), 700-16.

Interdepartmental Committee on Social and Economic Research, Guides to the official sources. No. 2. Census reports of Great Britain, 1801-1931 (London, 1951), 95-101.

W. H. Hubbard and K. R. Jarausch, 'Occupation and social structure in modern central Europe. Some reflections on coding professions', Quantum Information, 11 (1979), 10-19.

N. A. Humphreys, ed., Vital statistics. A memorial volume of selections from the reports and writings of William Farr (London, 1855).

C. Lee, British regional employment statistics, 1841-1971 (Cambridge, 1979)

G. B. Longstaff, 'Suggestions for the census of 1891', Journal of the Royal Statistical Society, 52 (1889), 436-67.

M. McDowell, 'William Farr and the study of occupational mortality', Population Trends, 31 (1983), 12-14.

D. Mills and J. Mills, 'Farms, farmers and farm workers in the nineteenth-century census enumerators' books: a Lincolnshire case study', The Local Historian, 27 (1977), 130-43.

D. R. Mills and K. Schürer, 'The enumeration process', in Local communities in the Victorian census enumerators' books, eds, D. R. Mills and K. Schürer (Oxford, 1996), 16-26.

D. R. Mills and K. Schürer, 'Employment and occupations', in Local communities in the Victorian census enumerators' books eds, D. R. Mills & K. Schürer, (Oxford, 1996), 136-60

R. J. Morris, 'Occupational coding: principles and examples', Historical Social Research, 15 (1991), 3-29.

Ministry of Labour, A dictionary of occupational terms based in the classification of occupations used in the census of population, 1921 (London, 1927).

W. Ogle, 'The alleged depopulation of the rural districts of England', Journal of the Royal Statistical Society, 52 (1889), 205-40.

B. Preston, Occupations of father and son in mid-Victorian Britain (Reading Geographical Papers No.56, Department of Geography, University of Reading, 1977).

G. Routh, Occupations of the people of Great Britain, 1801-1981: with a compendium of a paper by Charles Booth (Basingstoke, 1987).

K. Schürer, 'The 1891 census and local population studies', Local Population Studies, 47 (1991), 16-29.

K. Schürer, 'Understanding and coding the occupations of the past', in The use of occupations in historical analysis, eds, K. Schürer and H., Diederiks (St. Katharinen, 1993), 101-62.

R. H. Steckel, 'The quality of census data for historical inquiry: a research agenda', Social Science History, 15 (1990), 579-99.

S. Szreter, Fertility, class and gender in Britain, 1860-1940 (Cambridge, 1996).

P. M. Tillott, 'Sources of inaccuracy in the 1851 and 1861 Censuses', in Nineteenth-century society , ed., E. A. Wrigley,(Cambridge, 1972).

T. A. Welton, Statistical papers based on the census of England and Wales, 1851, and relating to the occupations of the people and the increase of population, 1841-51 (London, 1860).

M. Woollard, 'Creating a machine-readable version of the 1881 census of England and Wales', in C. Harvey and J. Press, Databases in Historical Research (London, 1996), 98-101.

M. Woollard, '"Shooting the nets": a note on the reliability of the 1881 census enumerators' books', Local Population Studies 59 (1997), 54-7.

TABLES

		Number			Percentage
Order		Report	Database	Diff.	Report	Database	Diff.
1	Government	144	148	-4	0.27	0.28	-0.01
2	Army/Navy	99	104	-5	0.18	0.19	-0.01
3	Professional	668	662	6	1.25	1.24	0.01
4	Domestic Serv.	4331	4409	-72	8.09	8.23	-0.14
5	Commercial	272	262	10	0.51	0.49	0.02
6	Transport	1351	1379	-28	2.52	2.57	-0.05
7	Agriculture	3747	3890	-143	7.00	7.26	-0.26
8	Animals	2139	2209	-70	3.99	4.12	-0.13
9	Books etc.	95	103	-8	0.18	0.19	-0.01
10	Machines etc.	116	112	4	0.22	0.21	0.01
11	Houses, furniture etc.	1760	1733	27	3.29	3.23	0.05
12	Carriages etc.	100	102	-2	0.19	0.19	0.00
13	Ships and Boats	164	155	9	0.31	0.29	0.02
14	Chemicals	58	58	0	0.11	0.11	0.00
15	Tobacco	47	50	-3	0.09	0.09	-0.01
16	Food and Lodging	1617	1559	58	3.02	2.91	0.11
17	Textiles	559	562	-3	1.04	1.05	-0.01
18	Dress	2316	2290	26	4.32	4.27	0.05
19	Animal Substances	38	37	1	0.07	0.07	0.00
20	Vegetable Substances	88	90	-2	0.16	0.17	0.00
21	Mineral Substances	1681	1684	-3	3.14	3.14	0.00
22	General/Unspecified	1052	1115	-63	1.96	2.08	-0.12
23	Refuse	10	11	-1	0.02	0.02	0.00
24	Unoccupied	31106	30859	247	58.08	57.59	0.49
	Total	53558	53583	-25

Table 1.

4		Published			Database
	Occupation	M	F	T	M	F	T	Diff.
054+	Coachman (Domestic)	83	0	83	32	0	32	51
055+	Gardener (Domestic)	208	0	208	211	2	213	-5
056+	Servant (Domestic)	47	3487	3534	95	3544	3639	-105
057+	Lodge etc. Keeper	0	0	0			0	0
058+	Inn, Hotel Servant	42	57	99	42	64	106	-7
059+	College, Club Servant	1	0	1	0	0	0	1
060+	Office Keeper	0	2	2	0	0	0	2
061+	Cook (Not Domestic.)	1	0	1	1	0	1	0
062+	Charwoman	0	211	211	1	217	218	-7
063+	Washing/Bathing Service	3	173	176	4	171	175	1
064+	Hospital/Instit. Service	8	7	15	7	17	24	-9
065+	Others in Service	1	0	1	1	0	1	0
	Total	394	3937	4331	394	4015	4409	-78

Table 2. Comparison of database and published report for those in Indoor Service.

5		Published			Database
		M	F	T	M	F	T	Diff.
100+	Farmer	1222	172	1394	1279	173	1452	-58
101+	Farmer's son etc.	557	0	557	619	4	623	-66
102+	Farm Bailiff	22	0	22	24	1	25	-3
103+	Agricultural labourer	1586	153	1739	1574	183	1757	-18
104+	Shepherd	15	0	15	15	0	15	0
106+	Land Drainage	0	0	0	0	0	0	0
107+	Ag. Machine	0	0	0	0	0	0	0
108+	Ag. Student	1	0	1	1	0	1	0
109+	Others in Agriculture.	0	0	0	0	0	0	0
110+	Woodman	0	0	0	0	0	0	0
111+	Nurseryman	11	1	12	8	1	9	3
112+	Gardener (Not Domestic)	7	0	7	7	1	8	-1
	Total	3421	326	3747	3527	363	3890	-143

Table 3. Comparison of database and published report for those in Agriculture.

		Published			Database
		M	F	T	M	F	T	Diff.
113+	Horse Proprietor	2	0	2	2	0	2	0
114+	Huntsman etc.	15	0	15	21	1	22	-7
115+	Veterinary	14	0	14	12	0	12	2
116+	Cattle, etc. Dealer	2	1	3	2	0	2	1
117+	Drover	0	0	0	0	0	0	0
118+	Game Keeper	2	0	2	2	0	2	0
119+	Dog, etc. Keeper	1	0	1	1	0	1	0
120+	Vermin Destroyer	0	0	0	0	0	0	0
121+	Fisherman	2102	0	2102	2164	4	2168	-66
122+	Knacker	0	0	0			0	0
	Total	2138	1	2139	2204	5	2209	-70

Table 4. Comparison of database and published report for those working with Animals.

		Published			Database
		M	F	T	M	F	T	Diff.
214+	Inn Keeper, Hotels	119	52	171	114	51	165	6
215+	Lodging House Keeper	44	280	324	45	279	324	0
216+	Coffee House Keeper	4	4	8	4	4	8	0
217+	Hop Merchant	0	0	0	0	0	0	0
218+	Maltster	7	0	7	7	0	7	0
219+	Brewer	23	0	23	26	0	26	-3
220+	Beerseller	0	0	0	0	1	1	-1
221+	Cellarman	3	0	3	3	0	3	0
222+	Wine and Spirits	16	0	16	15	0	15	1
223+	Milkseller	13	46	59	7	5	12	47
224+	Cheesemonger	0	0	0	0	0	0	0
225+	Butcher	141	5	146	136	6	142	4
226+	Provision Curer	15	30	45	14	28	42	3
227+	Poulterer	7	4	11	6	4	10	1
228+	Fishmonger	37	7	44	37	9	46	-2
229+	Corn etc. Dealer	26	6	32	21	5	26	6
230+	Corn Miller	82	0	82	85	0	85	-3
231+	Baker	237	15	252	238	16	254	-2
232+	Confectioner	8	49	57	6	48	54	3
233+	Greengrocer	14	24	38	12	17	29	9
234+	Mustard etc.	0	0	0	0	0	0	0
235+	Sugar Refiner	4	2	6	5	2	7	-1
236+	Grocer	224	64	288	226	60	286	2
237+	Ginger Beer, etc.	5	0	5	6	0	6	-1
238+	Others in food	0	0	0	3	9	12	-12
	Total	1029	588	1617	1016	544	1560	57

Table 5. Comparison of database and published report for those working in Food and Drink.

		Published			Database
		M	F	T	M	F	T	Diff.
402+	Manufacturer (undef.)	1	3	4	3	4	7	-3
403+	Contractor (undef.)	2	0	2	2	0	2	0
404+	General Labourer	719	54	773	720	65	785	-12
405+	Engine Driver	60	0	60	69	1	70	-10
406+	Artisan, Mechanic, etc.	22	0	22	32	2	34	-12
407+	Apprentice (undef.)	3	1	4	4	4	8	-4
408+	Factory Labourer	3	2	5	5	2	7	-2
409+	Machinist	4	0	4	3	10	13	-9
	Total	814	60	874	838	88	926	-52

Table 6.