Since the seventeenth century Scandinavian bureaucracies have created vast amounts of records on the individuals and families they administered. The most central are the censuses from the 1660's onwards and ministerial records that cover some areas even earlier. In addition, there are probate records and judicial material from roughly the same time, while records dealing with special groups like sailors or emigrants are of more recent origin. The quality and geographical coverage of these nominative sources vary over time and from country to country. The union between Sweden and Finland on the one hand, and the union between Denmark and Norway on the other, created different types of sources that continued to be specific within each of the twin countries even after the unions were dissolved in 1808 and 1814 respectively. This overview covers the national censuses, including those taken for only the towns or pre-censuses covering major parts of the countries. Censuses dealing with single towns, such as those taken annually for Bergen or Oslo, are not treated here. From 1875 to 1900, Norwegian ships sailing abroad should be included in the census, with special forms sent via the foreign services.

Historians and other researchers have for three decades made the material more available to research with the help of modern computer technology. This overview gives an outline of the history of the census material, how it was developed and constructed, what variables are included and specific problems that users should watch out for. The general impression is, however, that the Norwegian censuses are fairly complete and accurate, since the most severely flawed (1815) is believed to have an under-registration of less than three percent (Jahn 1929). Aggregate statistics from the censuses have been made available by the Norwegian Social Sciences Data Service and by Statistics Norway. ( Cf the web sites http://www.nsd.uib.no and http://www.ssb.no respectively.) An overview of the population size in the municipalities and provinces over time was printed as an appendix to the 1960 census (Statistics Norway 1964), and a more comprehensive historical overview of Norwegian statistial results on the national level in the seventh edition of Historical Statistics (Statistics Norway 1995). As a rule there are printed reports with aggregates from each census, with introduction and table headings in Norwegian and English or French. Two bilingual, printed catalogues of Norwegian official statistics covering the periods up to and after 1977 respectively are available (Statistics Norway 1978; Statistics Norway 1998).

The census manuscrips are kept by the National and Regional Archives. The original returns from 1801, 1865, 1870, 1891 etc are kept in the National Archives. Those in between (1875, 1885, 1900 etc), however, are kept regionally, here represented by the holdings of the Regional Archives of Tromsø. The 1900 manuscripts are in the parcels in the rightmost section, while the 1875 returns are in the protocols to the left of the parcels. With the exception of the 1891 census, all nominative censuses up until 1900 have been copied to microfilm, which is available from the National Archives, from the Church of Latter Day Saints in Salt Lake City and from many Norwegian libraries. While the master film is kept by the National Archives for the production of copies to users, the most comprehensive collection of films directly available to researchers is kept in the Regional Archive in Tromsø. In addition, certain parts have also been xeroxed, cf http://www.riksarkivet.no/arkivverket/kilder/medier/mikrofilm/mikrokort.html for an overview of the copies.

Archival aspects of the censuses are dealt with in three articles by (Isaksen 1988; Johannessen 1988; Martol 1988). Large parts of the nominative records are available in databases from HistLab (www.rhd.uit.no) and The Digital Archive (digitalarkivet.uib.no/). The formats recommended for the registration and electronic storage of the Norwegian censuses 1865-1910 are to be found on url www.rhd.uit.no/histform/ . Historical data bases in Scandinavia, some of them with censuses as a major type of source material, are dealt with in (Thorvaldsen 1998).

There was a shift from the use of ecclesiastic to modern administrative divisions and boundaries around 1875. The changes over time are few on the diocese, deanery and province level, while they are many and often drastic on the lower level of the parish or municipality. For an overview of old and new boundaries including sample maps, cf (Brosveet 1979; Statistics Norway 1980; Thorvaldsen 1997).

The source material

An overview of the European censuses, including the Norwegian ones can be found in (Goyer 1992). Regrettably, there is yet no comprehensive book on the Scandinavian censuses similar to Edward Higgs’ book on the British censuses (Higgs 1996), and neither do we have a comprehensive collection of research articles based on census material such as (Mills 1996). The Norwegian census type of source material may be classified into three groups: the male censuses, the statistical censuses and the complete nominative censuses. The latter category can be subdivided into censuses which have been made public, and those taken after the Statistics Act of 1907, whose use is restricted to statistical purposes only for one hundred years.

The male censuses of the 1660s and 1701

The scope of the first Norwegian/Danish census was restricted to less than half of the population - men and sometimes boys. Consequently these are called male censuses. This goes for the censuses from the early 1660s, which in the end came to comprise boys below 12 years, and the 1701 census which left out infants. The northernmost province (Finnmark) and the towns were left out in both instances, and large parts of the material from the latter census has gone missing for South-Eastern Norway and aa couple of parishes elsewhere. (It only exists for Rygge parish, Nedre Romerike and Nedenes bailiff’s districts, Odal parish and Larvik county.) Variables include information on farm names and tax value, names and social position; in 1701 also age and in Northern Norway separate lists for the Sami. The general quality of the first of the male censuses is better than the second, the main reason for this being the diligence of the Danish administrator Titus Bülche who kept corresponding with the priests until he received the lists he wanted. This 23 year old Inspector of Churches started to note census information on his first travel through Norway in 1662-3, only later obtaining the King’s permission to assemble this kind of information. His first ordinance in 1663 asked for information only about men who were twelve years or older, indicating the military purpose of the census-taking. However, in connection with his reminder to the priests in 1665, the scope was extended to all boys and men. On the other hand, users should be aware that this also resulted in many instances of double-counting, which is useful for source-critical purposes. Bülche’s appointment to the position of Mayor in Copenhagen in 1665, probably explains why he never finished his statistical extracts from the nominative records for more than selected places. The holdings of manuscripts, however, are virtually complete (Martol 1988). The material was used to estimate the population size by (Aschehoug 1890).

This is far from the case with the 1701 census, which was taken for military purposes, but not with the same diligence. Here the basic problem is under-enumeration, especially among infants, even if the whole male population ought to have been included (Lindstøl 1887). The partial nature of this source material suggests that we should call rather them pre-censuses. A searchable, digital version of the male census from the 1660s and the two provinces around Bergen from the 1701 male census is available via www.hist.uib.no/arkivverket/ Since the handwriting and ortography in sources from this period sometimes is difficult to copy, users should be aware that parts of this material have not been proofread.

Statistical censuses

The second type of census has been called "statistical". The reason for this is that - as a rule - we do not have census manuscripts with names and other information on each individual, only statistical aggregates about groups of individuals. The first census of this type was taken in 1769, and then with ten year intervals from 1815 to 1855. In principle the whole population was included, but we know about cases of severe under-enumeration, especially in 1815 and concerning the traveling part of the de jure population, which tended to be left out even if they formally were to be included with the household where they usually lived. For a number of parishes, draft nominative census manuscripts have been preserved, cf (Tranberg 1994) or url http://www.riksarkivet.no/arkivverket/kilder/medier/mikrofilm/mikrokort/ft1815-55.html for details. An overview of the contents of the statistical censuses can be found in table 1 (url http://www.rhd.uit.no/census/censustable.html ), which has been translated and adapted from (Dyrvik c1983). The aggregates from the 1815 to 1855 censuses are available on microfiche, and some tables were also printed together with results from later censuses (Bull 1882). This reference also contains a description in Norwegian of the procedures used for taking the censuses 1769 to 1865, made available at url http://www.rhd.uit.no/census/oldcensus.html .

1769

The statistical results from the 1769 census, being the first one covering the whole Norwegian population, has been published in book form by Statistics Norway (1980), including table headings and an introduction in English, cf url http://www.rhd.uit.no/rhd/census/ft1769.html . The original tables were transferred from Copenhagen in 1937 and are kept in the National Archives together with nominative lists for Bergen Fortress and six parishes in the northernmost diocese (Blix 1942). Some nominative and statistical list are also found in Trondheim Regional Archive (Johannessen 1992). It is hypothesized that the background for taking this census was the abolition of adscription and the ensuing uncertainty about what military consequences this could lead to (Momsen 1974). The Ministry of Finances in Copenhagen sent letters to the diocese administrators and bishops in Norway, informing them to distribute the enclosed forms to the town magistrates and rural priests. The latter should fill in the required numbers in the tables themselves, while this should be entrusted to two assistants in each town. There were two tables to be filled in for each parish or town, one grouping the population by age, the other by social status, both with categories for each sex. The de jure principle of counting was employed, but for security reasons most military persons were excluded. Their numbers are, however, known from other sources.

A source critical assessment of the returns for Eastern Norway (Akershus diocese) shows many inconsistencies between the two sets of tables. This is due to differences between the priests’ as to how the occupation forms should be filled in - no surprise when we take the minimal character of the instructions into consideration. After a detailed adjustment for the inconsistencies, it can be shown that the tables are fairly consistent (Sogner c1979), cf url http://www.rhd.uit.no/art/sogner.html . Sogner summarized her assessment in this way: "The 1769 census, however, is ill-famed. It is demonstrated that the bad reputation of the census goes back to opinions established by Professor G. C. Oeder for the census as a whole when, about 1770, he examined the figures themselves. The shortcomings pointed out are discussed for the Akershus figures and are not borne out. The conclusion then is that for the diocese in question, the figures seem reasonably sound, and their use justified." (p 519). Sogner stresses that the census tables for each district must be evaluated independently, so her conclusion for South-Eastern Norway may not be valid for the rest of the country.

The printed aggregates include tables on the level of the country, the diocese, the deanery, the parish and the town and each of the two northernmost provinces. Persons are grouped by sex, age, civil status and social group, but needless to say for a non-nominative census, with the exception of sex and age group the variables are not cross-tabulated. All ages above 48 were lumped together, the eight-year groupings used for those younger causing comparability problems against most later censuses. The problem of under-enumeration was enlarged by the taking of the census in mid-August, a season with many persons absent, especially sailors and fishermen. Whether this was the reason for the low male to female ratio along the southern coast or if it was due to migration to Holland, is still unknown. There is a similar problem in Northern Norway when comparing with the 1801 census and the number of baptisms and burials in the period in-between. The inconsistency may be due to in-migration to the northernmost region or its under-enumeration in 1769 (Drake 1969), cf url http://www.rhd.uit.no/art/drake.html . While Sogner accepts an in-migration to the northernmost diocese of 18-19000 people as likely, Drake does not. It follows that Drake was as critical of the 1769 census as others have been of the one from 1815. An overview, although somewhat outdated, of the demographic context can be found in (Gille 1949/50).

1815

There may be several reasons why this census, taken on April 30^th and the next few days, is the one with the lowest quality. The new Norwegian state had ceded from Denmark in 1814 and joined a personal union with Sweden, the administrative apparatus was reorganized and the government split with some ministers located in Christiania (Oslo), others in Stockholm. The state finances were miserable, a main reason being the Danish-Norwegian joint dept causing the next Danish census to be postponed until 1835. This time the parsons should be assisted by the country policemen, and the completed aggregate forms should be sent back to the Governments 4^th office. This census reports population totals for each town and parish by sex and occupation, in addition to the number of married people. Aggregate statistics from the 1815 census was published as part of the census work after 1855. Attempts at rectifying the under-enumeration in this census by comparing with other censuses and the vital registration, have been hampered by its lack of age information and data on any level below the parish.

The 1875 report from the Central Bureau of Statistics (Bull 1882) states that the 1815 census gives information about age groups, and this is substantiated by the offprints of the census forms, but not by the results published in 1816 (1815; 1816). According to his personal notes, Eilert Sundt has tried to distribute the population in 1815 across ten-year age groups (Manuscripts collection, the National Library). I cannot find that he has published these numbers anywhere or commented on his methodology.

1825

A special Statistical Office ("Tabellkontoret") was not established until 1817, but for long periods it was only staffed periodically with employees from other offices in the Ministry of Finances or the Ministry of the Interior to carry out special tasks such as the censuses. The non-permanent status and shifting locations of the Statistical Office shows its low priority within the central administration. However, the continued problems with manning and the economy did not stop the later censuses to have fewer problems with quality and under-enumeration than the one from 1815. Fortunately a suggestion in Parliament to reduce the costs of this census was not carried (Bull 1882).The variables were the same as in 1815, the major improvement being that we have forms for each address in the towns and each farm in the countryside. The timing was the first Sunday of Advent (November 27^th) and the next few days. Cf (Holst 1827) for further information.

For a nominative list of the inhabitants in Bardu, Troms in 1825, cf Bardu 1825

1835

The timing, contents and procedures connected with this census was similar to the previous one, except that the parsons could order their curates to assist them - on the other hand there was no longer any mention of the possibility to summon heads of households to report directly. For the first time, there were questions about about the blind, deaf and mentally deranged persons and grain and potato crops and livestock. It is unknown whether the introduction of such economic variables affected the quality of the census, since people were afraid the information could be used for taxation purposes. Background material and statistical results from this census were printed in a special publication (Statistics Norway 1838/9

. The census results were also used by (Schweigaard 1840). Information on occupations was only counted for the heads of household.

1845

From 1845 to 1875 all censuses were taken at the end of the census year. The reason for this rearrangement was that in 1839 it had been decided that vital registration of the number of married, born and dead, should be changed from following the ecclesiastic to the ordinary calender. Occupational information should be counted for all persons, and the number of people receiving poor relief should be specified together with data on lepers, while the categories concerned with mentally deranged persons were simplified. The results were published in (Statistics Norway 1847).

For a nominative list of the inhabitants in Bardu, Troms in 1845, cf Bardu 1845.

1855

The role of the parsons in the administration of the 1855 census was changed into one of supervision, while the practical enumeration should be carried out by the countryside policemen assisted by the teachers. The financing of the census was improved with a budget item for traveling expenses to the takers, which may have improved its general quality, especially in sparsely populated areas. The variables about lepers and paupers were dropped, and further specifications were made for the age groups and occupational categories. For results, cf (Statistics Norway 1857).

Nominative censuses that are publicly available

Censuses based on manuscripts containing information on identifiable information such as names have been taken in Norway in 1801 and then with approximately ten year intervals from 1865. The 1885 census only include the towns and the eastern part of Finnmark province, the next complete census being taken in 1891. There is a special census for the towns from 1870, and several local censuses have been taken, especially for Oslo and Bergen. Also, the rhythm had to be changed because of World War II, when the census was moved from 1940 until 1946. The last complete census was taken in 1980, in 1990 sampling was used with varying density according to the size of each municipality. Norway’s next - and maybe last - census has been postponed until 2001. The plan is to fetch household information from paper forms sent to all households, while using existing registers to assemble individual level data. In 1865 the de jure principle - just counting each individual where they had their formal residence was still used. In later censuses this was combined with the de facto principle, so that people should also be enumerated and marked especially where they were visiting or working temporarily.

The censuses up till and including 1900 can be used by anyone and for all purposes, except for the restriction that the data protection registrar has put on making identifiable individual level information about religion, ethnicity and illnesses public on the Internet. The Law on Statistics from 1907, however, stated that such material can only be used for statistical purposes, a clause that in 1986 was modified to "protect" the material for 100 years (Breivik 1988). Therefore, the 1910 census will not be made public until 2010, even though several researchers have been allowed upon application to use this and later censuses for their statistical investigations, and even data process the material. A summary overview of the information fields contained in these nominative censuses can be found at url http://www.rhd.uit.no/census/censustable.htm#Table2 . Large parts of the 1865, 1875 and 1900 censuses have been encoded for statistical purposes by HistLab, cf (Thorvaldsen 1995) for an overview in English. Sivert Langholm has summarized social history research on the Oslo region with the 1865 to 1900 censuses as the main source material (Langholm 1974; Langholm 1976).The latter article is in English.

1801

Norway’s first nominative, complete census was taken (together with Denmark, Iceland and the Duchies) on Sunday February 1^st and on the following Sundays if necessary. The rationale behind this administrative decision is unknown, but the existence of contemporary censuses in England and France suggests a mercantilist wish to assess the national population resources at the turn of the century. In the towns (10 % of the population) the census was administrated by the mayor ("magistraten") and carried out by the district registrars, in the countryside the parsons were assisted by the vergers and school-teachers. In addition, the priest could summon heads of household in order to receive the information if he deemed this appropriate. However, it is clear that the clergy did most of the enumeration personally by asking the heads of household after Sunday service.

The scope of the census followed the de jure principle, so military persons should be included (in their barracks) as well as foreigners if they were residents. It is unclear what over- and under-enumeration was caused by the intended exclusion of non-citizens and inclusion of resident foreigners. (The latter is an instance of de facto registration, which is not in accordance with a modern interpretation of the de jure principle.) In some parishes the Sami are enumerated separately, but the absence of ethnic information in other places warns us about confusion as to where they were (supposed to be) enumerated and their possible under-enumeration. Also, the extensive, seasonal fisheries taking place in February involved much temporary migration and has probably led to some under-enumeration, while on the other hand the merchant fleet is less active during the winter months. The manuscripts for the parish of Holt-Dypvåg on the coast of Southern Norway has gone missing sometime after the 1820s, so for this locality we only know the most basic figures. Also, there is no information on a small number of farms, which were probably inhabited at the time. This may be explained by the seeming lack of a clear order of counting in the countryside, while the town registrars were supposed to walk from house to house.

The picture below shows the original 1801 census manuscripts in the leather bound protocols on the right, together with 17th and 18th century census material in the National Archives in Oslo.

The original 1801 Census Manuscripts

The over-representation of ages ending in zero and five shows us the approximate nature of this infor- mation. The data on the number of times someone had been married, on the other hand, is a special feature in this census. More unclear is the contents of the household definition employed by the census takers, except that they probably included persons living in the same house. Differing practice as to whether lodgers and retired parents were counted together with other residents, should be interpreted as unclear definitions rather than to mirror differences in household status. Also, the information about occupation typically is vague, often reflecting a person’s status rather than by what trade he made his living, and leaving out many of the secondary occupations. This means, for instance, that the category of fishermen is undercounted, since they frequently were listed as farmers or cottars. Even so, there are 40000 different text strings representing the occupations in the census, but this figure is of course inflated by inconsistent spelling.

In the coded edition of the census made available on the Internet, this is reduced to 300 different occupation. For further information see the introduction to the revised edition of the census by Ståle Dyrvik (cf below). Unfortunately, there is no data on birth place in the 1801 census. The printed edition includes maps comparing the old parish divisions used in 1801 with modern administrative borders according to municipalities.

During the 1970s the nominative lists from 1801 were data processed by the Historical Institute at the University of Bergen in cooperation with Statistics Norway and the National Archives. This has resulted in a searchable edition available on the Internet at url http://www.hist.uib.no/cgi-win/websys.exe?E, and a new and a revised printed edition of the statistical results (Statistics Norway 1980). The English version of the introduction to the publication (Dyrvik 1980) is found at url http://www.hist.uib.no/1801en.html. For an example of research using the 1801 census and ministerial records to study social mortality differentials, cf (Engelsen 1983) or url http://www.rhd.uit.no/Ht/ht62.html#2169.

1865

After half a century of statistical census taking only, the census taken in the beginning of January 1866, was a novelty in more than one way. Most important, the name and other information about each inhabitant was noted for the first time since 1801. (An interesting exception is the parish of Tana in Finnmark province, where the names are left out for persons subordinate to the head of household.) Also, the self-enumeration principle was introduced for the first time in Norway. In the towns, the house owners were requested to fill in the census forms themselves, forms which were to be distributed, collected, checked and if necessary completed by the district registrars. The forms should be signed by the enumerator, and for the town of Tromsø this field has been used to study who the census takers were. Only in two cases were the forms signed by the house owner, as a rule they were filled in and signed by some relative, some neighbor or someone else living in the house (Drake 1991). This is in contrast to the report in the printed results from the census, where it is maintained that the house owner only had to be assisted in a few cases (Statistics Norway 1868-69). In the countryside, the police assisted by the schoolteacher were still in charge of filling in the forms. At least formally, however, the ministers should direct the enumeration and control the census manuscripts before they were returned to the Department of the Interior. Complaints from the priests and the teacher about the increasing work load involved in census taking were not heeded by the authorities, but at least the teachers were paid to do the census work (Tranberg 1994). Since they knew the households better than other officials, the Ministry felt that their assistance could not be dispensed with. In order to cover the inhabitants of all farms, each enumerator received a copy of the printed farm tax assessment list for his district. These lists are organized topographically, and could be followed when going from farm to farm. The 1838 version of the tax lists was used until a new edition was printed during the late 1880's. Since cottars’ farms were not included in the lists and often are less centrally located, maybe these were more likely to be under-enumerated. There are indications that the 1865 census is somewhat less complete than later censuses, but the main reason for this may be that it was the last to include only the de jure part of the population, so it was more likely to leave out temporary migrants.

The 1865 census was the first in Norway to give place of birth for each person listed. The parish was often noted instead of the municipality, and users should be warned that this may be the parish that a municipality belonged to at the time when the person was born, but which has later been changed (Thorvaldsen 1995). (Cf url http://www.rhd.uit.no/nhdc/summary.html for a summary of this dissertation which contains a number of source critical points concerning the 1865, 1875 and 1900 censuses.) For the last time age was given instead of birth year, which may have led to errors because people reported their present age rather than the one they would reach in 1866 as instructed. As could be expected, round ages are over-represented, a phenomenon which gets to be less frequent in later censuses (Thorvaldsen c1996). Family and household information should be given, but it has been found that in many cases a formal position such as "servant" was given instead of actual family relations. Occupation was lumped into the same field as data on household status, and in addition to the preference for formal titles ("cottar") rather than work functions, there is a widespread under-registration of secondary occupations. For instance the 1865 census is known to vastly understate the number of fishermen. In Northern Norway and in some other places, the ethnicity of each individual should be specified, even if the forms provided no special field for this piece of information. In many cases this created problems, since some families were split and the census manuscript instead organized according to ethnic belonging. The Histform standard requires that this information should be placed in its proper field, and this is the case for the part of the 1865 census entered by HistLab.

The original manuscripts are kept in bound volumes in the National Archives, with xeroxed copies in some regional archives. The exception is the parish of Gol in Eastern Norway and some summary lists for Oslo which have gone missing..The instructions to the census takers and sample census forms are available on url http://www.rhd.uit.no/census/ft1865.html There are slightly different forms for towns and the countryside, most notable the address fields where farm number are left out in towns. Statistical results for the population is available in (Statistics Norway 1868-69), while agricultural figures are reported in (Statistics Norway 1869).

1870

This census covered only the towns, where the authorities felt they needed updated information because of rapid urbanization. From 1835 to 1920 the Province Governors ("Amtmenn") sent five-annual reports to the statistical authorities. Late in 1870 they were ordered to include an urban census in their next report due. The census should be taken and financed by the administrators of municipalities with formally recognized urban settlements (byer eller ladesteder). The Department of the Interior included a sample census form, but since these were to be printed by each municipality, the forms used are somewhat different from place to place (Isaksen 1988).

One town abstained from taking this census altogether, and the manuscripts for several others have gone missing, so the records for ten communities cannot be found in the 1870 collection in the National Archives. This material has not been microfilmed Few if any of these census returns have ever been computerized or used in research, and it is not known if any statistical results were published.

1875

While the nominative basis of the 1865 census was a great leap forward, the methodological changes introduced in census undertaken in the beginning of January 1876 meant a significant improvement in quality. This resulted both from the generally improving standards of education in Norway, and more specifically from the organizing of a special administrative agency for the production of statistics called Det statistiske Centralbureau (The Central Bureau for Statistics) in 1876. This had been prepared by employing more specialized personnel during the previous years. Most notably is the introduction of the explisit distinction between the de jure and the de facto population, with special fields for noting the whereabouts of absent people and the origins of temporary residents. However, the information about absent persons were noted on special lines at the bottom of each form, which sometimes makes it difficult to see which family or household they belong to, and creates extra problems when coding family relations. The standard specifications for computerized versions of Norwegian censuses, therefore, states that these persons are to be moved to where they belong among the persons present if possible. (Nygaard 1995)Spot tests indicate that the 1875 census is even more complete than the previous one. For completeness and adherence to the de jure principle, census forms and instructions were also sent to Norwegian ships abroad via diplomatic channels, the forms being only marginally different from those used ashore. These despatches also contained letters with brief explanations in French and English, cf url http://www.rhd.uit.no/census/ships75.html.

Another change was the reporting of birth year instead of age for each person. The intention behind this reform was probably to eliminate the confusion arising from the choice between reporting either the previous or the next birthday. Spot tests again indicate some improvement in correspondence with baptismal records, but since the census takers would ask for age and compute the year of birth, misreporting is still quite frequent although discrepancies of more than one year are seldom found. The improvement in the reporting of occupations was more significant, since the one field about both household and occupation was now split in two. Also, the census takes were explicitly told to report also extra sources of income. Not only did this result in the mentioning of many adjunct jobs, it also became more customary to note the actual work function, and not only formal titles such as "cottar" or "lodger". Even so the fishermen category is not complete, "primarily because grown-up sons living with their parents on the farms, and many servants who participated in the fisheries, where not noted as fishermen." (Solhaug 1976) In the northernmost provinces, a special form was used with an extra column for ethnicity, while the margin still had to be used to note data on languages spoken. Instead of a general ethnic label, now the ethnicity of each of the parents should be noted. Therefore, we can easily measure the degree of out-marriage in ethnically mixed areas. Of course, when one of the parents was of mixed ethnic origin, the category "mixed" still had to be used. In essence, however, this reform probably led to more weight on the genetic rather than the cultural aspects of ethnicity. And even if spot tests indicate that the information on ethnicity as a rule is accurate, we should remember that a single person with different ethnicity from the rest of the household could easily be amalgamated. The information about languages spoken is often rich, indicating that many people were trilungual, speaking both Sami, Finnish and Norwegian. But we ought to be suspicious of information where ethnicity is not in accordance with language. For instance, if the form reads that Sami people spoke Norwegian, it probably means that they spoke Norwegian in addition to Sami (Thorvaldsen c1996).

The main results from this census are reported in (Statistics Norway 1878-81; Statistics Norway 1882). A summary of the proceedings in Norwegian can be found at url http://www.rhd.uit.no/census/ft1875.html . The census manuscripts from major parts of the country have been computerized by the Norwegian Historical Data Center, most notably Oslo, Bergen, Trondheim and Northern Norway. Now a 2 % household sample of households from the rest of Norway is being digitized. For an overview of municipalities covered, cf (RHD 1997) or url http://www.rhd.uit.no/datakilder.html.

1885

Also in 1885 the census was intended to cover only the towns where rapid urbanization had led to the need for an update since the next national census had been postponed until 1891 in order to be synchronized with the censuses taken in many other countries. (After the establishment of the special Statistical Bureau, Norway participated in the international statistical conferences.) Unlike in 1870, the printing of the census forms was financed and done by the Statistical Bureau, while local authorities were responsible for the carrying out of the enumeration, including its costs. However, the scope of this census was extended to cover five rural municipalities in the easternmost part of the northernmost province Finnmark, districts close to the border with Russia. This was allegedly because the authorities wanted to map the large Finnish population, which was (without reason) viewed as a security risk because of their ties with Finland, ruled as semi-independent grand-duchy under Russia until 1917. The information found is similar to that given in 1875, but in addition there is a field with information about employers. Only the manuscripts from a few towns covered in the 1885 census have been computerized, but there is a printed publication giving aggregate statistics (Statistics Norway 1887). Like for the 1891 and 1900 censuses, spot tests indicate that a number of manuscripts have gone missing.

The 1865 Protocols

1891

With a growing number of inhabitants and variables to handle, the census authorities were on the look-out for ways to streamline the process of taking a census. While the 1865 census forms were continuous and the later censuses used on form for each domicile, the 1891 census for the first time introduced one sheet for each individual. Since the forms can be sorted by the contents of each variable, this probably speeded up the enumeration process. Later on the archives had to resort the material according to topography. However, it is a hassle to copy all the individual sheets and the accompanying ones for each domicile, so this census is the only one up until the 20^th century which was never microfilmed. (The picture above contrasts the 1865 protocols on the right with the large number of boxes containing the individual forms for the whole country in the National Archives in Oslo.) Certain parts, like Bergen, have been copied, but otherwise the non-existence of backup copies has hitherto blocked the computerization of the 1891 census. This is a problem since the original is rather inaccessible, both because it is archived centrally in the National Archives, and because going through all the sheets involves a lot of page-turning. So far, it has not been discovered that any significant parts of the census has gone missing because of extra work involved in arranging this kind of material. However, the agricultural part of the census was destroyed in a fire in 1939, so only statistical results are available.

From 1891 on, the forms used in towns contained special fields in order to note information about each flat in apartment buildings, and the number of variables about domiciles was increased considerably. This source may, therefore, be called the first combined census of people and habitation. On the personal forms, two new variables were introduced. First, in addition to a field specifying the person’s own occupation, an extra field gives the occupation of the primary breadwinner in the household. This makes it easier both to be more specific about occupations and relationships within the family or household. Also, an appendix to the instructions list occupational titles that need further specification. For instance the census takers were ordered to note what kind of work a mechanic did, and where he was employed. (A similar list instructed them to add the name of the province to any ambiguous municipality name.) Second, it should be noted in a separate field whether or not the two spouses were related by blood ties, probably a concession to researchers interested in issues of heritability. In several of the fields, the census takes could underline keywords or abbreviations instead of or in addition to writing an ordinary string of letters. For instance in the marital status field there were preprinted terms like ‘Unmarried’, ‘Widowed’, etc, and in the occupational field the words "Self-employed" or "Employed" should be underlined. The person’s own ethnicity should be given, and there was a special field for language in the parts of Norway that are ethnically mixed.

The statistical results from the 1891 census are available in several volumes, cf (Statistics Norway 1894-98).

1900

This was the first Norwegian census to be tabulated mechanically, using the kind of equipment with punched cards and electromechanic counters that Hollerith had developed for the US census of 1890. In addition to saving work at the time, this meant the reintroduction of forms for each domicile and the dropping of the individual sheets since there was no longer any need to sort individuals manually. So far the 1900 census is the latest one to have been made public (in 1960) and transferred onto microfilm by the Church of Latter Day Saints.

Even if the layout of the forms is very different from 1891, the contents of the fields is quite similar. In both censuses there is a special form for information about each apartment in the towns. There are no questions for additional information, except for the specification of birth date for children two years old or younger. On the other hand, the questions about the occupation of the primary breadwinner and blood ties between spouses were dropped. Agricultural information was taken, but has survived for the towns only, since the remainder burnt in 1939. In the relevant districts, the forms contained extra fields, one about ethnicity and one about language. When the forms were processed in the Central Bureau of Statistics, however, ethnicity information was crossed out and corrected if not in harmony with data on the person’s language. Thus, a cultural definition of ethnicity prevails in the statistical results. As a rule the ethnicity data originally noted is legible and has been transferred into the machine readable version which exists for the whole country.

The 1900 census is available in machine readable formats for the whole country from the Digital Archive and HistLab - that is with the exception of a number of lists which cannot be found in the regional archives (mainly for urban areas). The statistical results were published in six volumes plus a "Main Overview" (Hovedoversigt) also containing the instructions to the census takers and copies of the forms used (Statistics Norway 1906). The instructions are available at url http://www.rhd.uit.no/census/ft1900.htm together with examples of the census forms.

Nominative censuses that may only be used for statistical research

The Statistics Act of 1907 stated that source material assembled under its jurisdiction, may only be used for statistical purposes. Since the law does not apply retroactively, censuses taken earlier were made public by the National Archivist after sixty years. The Norwegian Parliament, however, forgot to limit this clause in time, and this was not done until 1984 when Statistics Norway extended the restricted period to one hundred years. Therefore, the 1910 census will not be made public until 2010, fifty years after the census from 1900. Under an agreement between the National Archives, the Data Protection Registrar and the Norwegian Historical Data Center, the 1910 and 1920 censuses may be transferred to machine readable media, but only used by researchers provided they sign a letter promising not to disclose any identifiable information. The same kind of letter must be signed by researchers wanting to use any censuses from the 20^th century. Material up to 1950 are kept by the National or District Archives in paper formats, while later censuses also reside on computer media controlled by Statistics Norway. Anonymous copies of the latter data sets may be ordered by researchers for restricted use.

1910

The chairman (ordføreen) was in charge of taking this census in the rural municipalities. Since a modernized version of the electro-mechanical equipment could be used for the 1910 census, these manuscripts are only moderately different from 1900. The question about date of birth was made general, not only those below two but everyone was to give their birth-date. The biggest change, however, was the introduction of a special form for returned emigrants, with several detailed data fields: When emigrated, when returned, where the emigrant had lived, and what occupation he or she had at the temporary place of residence. It is assumed that the number of returned emigrants in this census was underestimated. The scope of this census was somewhat reduced, since the ship questionnaires were only sent to ships (Norwegian and foreign) in domestic ports or waters. Thus, crews and passengers on board Norwegian vessels in international waters or foreign harbours were not enumerated. Unlike in 1900, this census was not coordinated with the collection of agricultural statistics, but instead with a special enumeration of artisans and the crafts.

The statistical results from the 1920 census were printed in x volumes xx, with instructions and manuscript forms printed in the main volume - Hovedoversigt (Statistics Norway 1916). The manuscripts have been data processed for the town of Haugesund, but can only be used for statistical purposes and with special permission.

1920

When preparing the 1920 census, it was made clear that the automatic enumeration system had to be renewed, but this could not be financed because of the post-war recession. Therefore, the Statistical Central Bureau had to revert to the system of individual census forms used in 1890, with the extra complicating factor that fields had to be printed and filled out on both sides of each sheet. The extra copying involved when reproducing double-sided safety copies of the more than two million sheets, makes this source material the least available and most cumbersome and expensive to data process of all the Norwegian censuses. As for 1910, only the returns for Haugesund have been data processed.

The number of questions about returned emigrants was reduced in 1920, keeping only those about xxx. On the other hand, their number is supposed to have been more accurately assessed. New questions for the whole population concerned marital history and for the women fertility history.

The statistical results were published in thirteen volumes, with a comprehensive section on the history of Norwegian dwellings. Instructions and census forms were reprinted in the last volume.

1930

Fortunately, in 1930 the system with a common form for the persons living in each domicile was reintroduced. Since mass emigration to the US was no longer a threat, the questions about returned emigrants were now dropped, and instead there were question about income and unemployment.

Aggregates were published in ten volumes, with forms and instructions in the last one.

1946

The German invasion on the 9^th of April disrupted the plans for a census in 1940, but the war also created special needs to enumerate the population after its end. Norway’s next census, therefore was taken on December 3^rd 1946, the census forms being similar to those used in 1930. The household definition was based on units of residence, so lodgers having their own room was considered an independent household regardless of whether they had meals with the main household or not. Domestic helpers, however, were included in the household where they worked.

Statistical results were printed in six volumes, which do not contain the census forms or instruction. There is a short introduction in volume 1, and introductions about occupations and dwellings in the relevant volumes. Some results were published together with the statistics from the 1950 census.

1950

This census was also used to update the population registers started during the war. The household definitions were complicated with only lodgers having no meals with the others being tabulated separately, and a special board household category. New questions were introduced about use of the Sami language and educational attainment, people being asked to specify any certificate or degree requiring an education beyond five months. The marital history items included a question about the number of children born alive in the present marriage.

The aggregates were published in ten volumes, with census forms and instructions towards the end of volume 1. The secondary language used in prefaces and table heading was changed from French to English from this time on. There is a special computerized version of the results on the sub-municipal or school district level for this and the subsequent censuses from the Norwegian Social Science Data Services, cf url http://www.uib.no/nsd (Kretsdatabanken).

Modern, computerized, non-public censuses

Since 1960, Statistics Norway has used computers to produce statistical results from its censuses. (If a punched card version of the 1950 census was ever produced, we have little information about its formats or present whereabouts.) The computer versions that can be ordered by researchers are all anonymized and contain encoded values only. There is a bibliography of publications with results from the postwar censuses in (Statistics Norway 1999).

1960

Not much was changed in the 1960 census, except that relationships from all persons to both head of household and head of family was made more specific, with the old household definition of someone eating at least dinner in the same unit being reintroduced. This was the last Norwegian census to take all its information from questionnaires, since no central population register was available at the time. In fact, the 1960 census was the basis for creating such a register. The school limit of vocational training was extended from five months to one year, and the school enrollment variable indicated the educational activities of persons above primary school age. Place of birth was more precisely defined, namely as the mother’s residence at the time. The question about the use of the Sami language was dropped.

The aggregated results were published in nine volumes, including a list of unpublished tables in each volume. (These may still be unavailable to researchers.) The ISCO 1958 standard was used to tabulate occupations. The data documentation, including that for the computer tapes, is in Norwegian only, but table headings are in English as usual. Instructions and copies of census forms were not reproduced as part of the publications, as for the 1946 census, these are available on request from Statistics Norway only. As part of the 1980 census publications, a special volume compares the procedure for the 1960, 1970 and 1980 censuses, cf below. In order to perform the record linkage necessary, personal ID numbers were introduced into the 1960 census retrospectively.

1970

This census for the first time took some data, such as age and marital status, from the population registers, but the main source of information was still the individual and household forms filled in by self-enumeration. For the last time census takers were used to deliver and fetch the forms. Names, addresses and marital status was taken from the central population register and preprinted on the forms. In addition, a housing census was carried out simultaneously. The census distinguishes between group quarters, private households and non-resident individuals. A change in definition of household entailed that in 1970 there was only one household in each private dwelling, while in earlier censuses there could be two or more depending on who had their meals together. Any person without a permanent residence was considered a separate household. New variables were place of work, hours worked and commuting, the latter derived from the difference between place of residence and work. The citizenship variable and the de facto principle were dropped. But on the whole the 1960 and 1970 censuses were quite similar.

An innovation was that the quality of this census was evaluated against a post enumeration sample of persons aged more than fifteen, indicating an underestimation of the number of economically active people and number of people with vocational training. The results were published in six volumes, one volume containing a copy of the questionnaire. In addition there is one volume for each municipality. Some unpublished tables and computer tapes are available under certain conditions.

1980

The use of existing population registers in this census was extended to be the basis for all information on persons aged below sixteen, while the rest of the population gave information in personal and housing forms which they for the first time received and sent in by mail. The oldest person in the dwelling was responsible for filling in the housing form. The educational variables were taken from the Register of Education, and supplemented with information on exams taken abroad from the questionnaire. The household definition distinguished between group quarters and private dwellings, the latter being defined as one household unit, just like in 1970. A person not belonging to the two categories above, was called a one-person-family. Economic activity was defined both according to work during the last year and the last week, but confusion about the latter definition may have reduced the data quality. (Otherwise a post-enumeration survey indicated high data quality.) The new income variable was based on tax assessment. Means of transportation to work, duration of commute and number of weekly trips was asked for the week preceding the census. The place of birth question was dropped.

The statistical results were published in five volumes with additional printed reports for each municipality and province. There are English translations of classifications, table headings and summaries. The volume called "Landssammendrag" (Country Summary), contains copies of the questionnaires and instructions in Norwegian. Special reports deal with methodological issues.

Volume IV contains a comparison of main figures from the 1960, 1970 and 1980 censuses with text in Norwegian and English (Statistics Norway 1986). These aggregates are based on the totality of persons present in at least one of the three censuses. Information about individuals present in two or all three censuses has been linked using the unique personal identity numbers. Also, the variables have been recoded so as to maximize comparability between the three censuses, especially with respect to occupations and commuting. This makes it possible to group the population according to the status held by the same individual on the same variable in different years. For instance 709 persons were divorced in 1960,remarried in 1970 and widow(er)s in 1980. A special report documents the comparability and quality of this edition of the 1960 to 1980 censuses (Vassenden 1987). Computer tapes from the 1980 are available to researchers on certain conditions, but the same may not hold true for the linked 1960 to 1980 data sets. However, Statistics Norway offers to create tables according to specifications from users.

1990

The 1990 census took some data from existing population registers and only sent personal questionnaires to selected individuals, sample density depending on the size of the population in each municipality. In all municipalities with a population of less then 6000 inhabitants a full census was taken, while in the bigger ones sample densities varied between 20 % and 8.3 %. Sampled individuals also gave data on the other members of their households. Demographic information, including all variables for individuals up to 16 years old, was taken from existing registers, while these were combined on the individual level with questionnaire data to assemble occupational statistics. Household information was mostly taken from the forms sent out. It was estimated than the sampling procedures reduced the costs of taking the census by some 40 %. Some results are marked as unreliable or not reported at all because they are based on small samples.

The household definition was the same as in 1980, so soldiers and unmarried students are not counted in their camps or dormitories, but where they lived before starting service or studies. The main sets of variables in this census deal with housing, economic activity and commuting. The question about religion was dropped, showing the increased importance of protecting personal privacy.

The publication plan was similar to that for 1980, with the questionnaires and a description of procedures in a special bilingual volume entitled Documentation and Main Figures (Statistics Norway 1999), while other aspects of the census, such as sampling and the encoding of occupations are documented in separate volumes. There is one volume with figures for each municipality or each urban district in Oslo.

2001

The census planned for the turn of the millennium will take place in 2001 and will be based on data about the whole population. This will be the first Norwegian population census entirely based on existing registers, since the questionnaires sent out to all households will assemble housing information only. Information will taken from the central population register, the register of employers and employees, the register of income and wealth, the register of education, the company register and the register of properties, addresses and dwellings. Statistics Norway works to update the population registers in order to make register based censuses into an option giving results at the same quality level as one completely based on traditional methods. At present this involves establishing longitudinal information on variables such as type of work and education taken abroad. Where possible, the 2001 census will comply with the "Recommendations for the 2000 censuses of population and housing in the ECE Region", UN/ECE Statistical Standards and Studies - No. 49

In order to prepare for registry based housing censuses as well, this part of the 2001 census will be coordinated with un update of the central registry of properties, addresses and dwellings (GAB - Gate, Adresse, Boligregistret). A unique dwelling number now exists only for separate houses, but will be introduced also for flats etc in the GAB-register, the population register and the 2001 housing census forms. The 2001 census project has its homepage at url http://www.ssb.no/emner/02/01/fob2001/.

It is claimed that Denmark did the world’s first register based census in 1980. This is true only if we add the terms modern or computer based, since the first Swedish censuses from 1860 were based on extracts from the catechismal ministerial records.

{- To be expanded}

References