Chapter 1 Problems of the population historian (pp 1-6 on the censuses 1801-1865)
History, traditionally viewed, is a literary exercise: 'charm and style', (1) are the hall marks of good historical scholarship. Population history, however, is more for the numerate than the literate, since numbers, not words, are the basic although, of course, not the only material. Without adequate quantitative evidence the demographic historian should not pick up his pen. If he does he will rarely be credible. He might well, of course, be more readable and this is his dilemma: the more statistics he has, the more authoritative his writing becomes and, not infrequently, the more unreadable. But to eschew statistics, to write in the traditional manner, is to contribute little or nothing to the understanding of our demographic past.
It is of course one thing to demand quantitative evidence, quite another to produce it. Population statistics are rarely, if ever, perfect and become less so the further back in time we probe. Yet it is because the success of population history depends so much on the representativeness and accuracy of the statistical material that the demographic historian must needs spend much of his time and energy in assessing its worth.
The value of any statistics depends upon the skill with which they are collected, processed and analysed. Errors are likely to be made during each of these stages and if not spotted may be compounded possibly with horrendous results.
In the case of population statistics the primary material from which they are derived is collected in two ways: by a census or enumeration at a point in time and by a recording or registration process through time. The demographer takes a census in order to draw up a balance sheet of the various characteristics of the individuals making up the population: age, sex, marital status, occupation, place of residence and so on. The registration process is used to record, as they happen, the events which bring about population change; births, marriages and deaths being the important ones for most purposes.
Censuses covering all, or almost all, a western country's population were not taken until the eighteenth century, the first being in Iceland in 1703. This was followed by Sweden in 1749, where a unique system was inaugurated which combined the registration and enumeration processes. The kingdom of Denmark-Norway carried out its first census in 1769, the United States in 1790, England and France in 1801. Most other European countries did not begin to take regular censuses until later in the nineteenth century. (2)
These first censuses were a great advance. Inevitably, however, they leave much to be desired. Neither the people who planned them nor those who actually carried them out had any experience of this kind of work. They tended to ask only a limited range of questions and even these were frequently ambiguous. Enumerators were sometimes recruited and paid in ways not likely to enhance their efficiency. The Irish census of 1831, for example, probably resulted in an over-count partly because it was spread over a considerable period of time and partly because the enumerators thought that 'their payment would be in proportion to the number of people included in the returns '. (3) Under, rather than over-enumeration was, however, the main problem of the early census takers. The populace was often suspicious and uncooperative. Poor communications made the task of the enumerators arduous and sometimes even dangerous. Children, particularly those under five years of age, were frequently missed. When ages were asked digital preferences were often very marked. The ages commonly given ended in either a nought or some other even number.
Eight censuses were taken in Norway between 1769 and 1865: two in these years and the others in 1801, 1815, 1825, 1835, 1845, and 1855. (4) The best were undoubtedly those of 1801 and 1865 as they, were the only ones where the enumerators had to record the name of each inhabitant together with his or her age, sex, marital status, occupation, position within the household and address.
For the 1801 census considerable pains were taken to see that the instructions were carried out properly, and completed sample forms were sent to each enumerator before the count was taken. The clergy acted as enumerators in the countrysides, the magistrates in the towns. On the question of marital status not only was the enumerator required to find out whether the person was single, married or widowed, but also whether he or she had been married more than once and if so how many times. People temporarily away from the household in which they normally lived, no matter if they were in some other part of the country or overseas, were to appear in the census as if they had been in their normal residence on census day, 1 February.
In the towns the magistrates, accompanied by local registrars, were to visit the heads of each household and obtain the requisite information directly from them. In the countryside the priests were not obliged to visit the different households, but were instead to announce from the pulpit a time and a place at which they would meet the heads of households. Priests were assisted in filling out the census forms by local school teachers and parish precentors.
It was obvious to the organisers of the census that not all these meetings could take place on the one day and that unless care was taken this might well spoil the census. They therefore warned the enumerators specifically to omit from the census children born after 1 February and to make sure not to omit those people who had died after that day. It was also realised that the returning of ages might easily lead to confusion. To avoid this two illustrations of how ages should be recorded were given to guide the enumerators. Thus a newborn child was to be entered as aged one year; a man in his 26th year was to appear as 26, not 25 (i.e. his last birthday). (5)
Despite these precautions, the general failings of census-taking in relatively primitive societies appear, although one supposes they were less evident than they might have been. A check of some burial registers for the years immediately after 1801 reveals the deaths of people not recorded in the census. (6) One finds, too, digital preferences leading to the bunching of the population around the ages 10, 20, 30, 40, etc. A large number of men in the valley of Hallingdal chose to return themselves as 36 years of age, possibly because liability for military service ended at that age! (7) In spite of these shortcomings, the census of 1801 was far superior to any taken outside Scandinavia either by that date or for many years to come, and the range of detail it supplies allows us to dissect the Norwegian society of the time with considerable precision. It is a great pity that a census of this type was not to be taken again until 1865.
The remaining six censuses were much less satisfactory. None of them were nominative. The enumerators using tally sheets were merely required to return population totals broken down by age, sex, marital status, and occupation. Because no names were required the task of assessing the reliability of these censuses is very difficult. One can fairly safely assume, however, that they covered the population less completely than did the censuses of 1801 and 1865. The two least satisfactory of this non-nominative group of censuses were those of 1769 and 1815. The first of these was taken in August, a month when many people were likely to be away either on the fishing grounds or tending their animals on the high mountain pastures. There appears also to have been quite a widespread fear that the census was a prelude to higher taxation. It was, after all, the first census and it came shortly after a particularly onerous poll tax had been introduced in 1762. (8) This fear, we imagine, must have led to some under-enumeration.
The census of 1815 also appears to have suffered from its timing - just after the Napoleonic wars had ended and during the early days of Norwegian independence. The instructions sent to the enumerators were much less detailed than in 1801 and were sometimes misleading. (9) No particular effort appears to have been made to ensure an accurate count.
For the censuses Of 1845 and 185 5 the enumerators were provided with a list of all the farms in their district. This had been drawn up for rating purposes in 1839 and obviously reduced the likelihood of entire households being overlooked. The officials in the department of the interior responsible for the publication of these two censuses used the list to cheek the work of the censustakers. Their comment on the census of 1845 was that the returns showed 'care and accuracy '. (10) Their only comment on that of 1855 was that in the agricultural part of the census, the amount of seed sown and the number of animals on the farms appeared too low. (11) This agricultural section had first been attached to the population census in 1835 and was repeated in 1845 and 1855. (12)
Despite these strictures, leading figures among those who have worked on the early Norwegian censuses have not been sparing in their praise. Eilert Sundt remarked that the more he worked on them the stronger his impression of their reliability became. (13) When Anders Kiær, the first director of the central statistical bureau in Norway, came to revise the census totals he estimated that the most inaccurate of them all, that of 1815, was no more than 3 per cent deficient. (14) Gunnar Jahn, another notable director of the bureau, believes that the deficiency was probably even less than this modest figure. (15)
1. Douglass C. North, 'The state of economic history', American Economic Review, 55, no. 2 (May 1965), 86.
2. Excepting the 1665 census of New France. A. M. Carr-Saunders, World population: past growth and present trends (Oxford, 1936), pp. 6-8. T. Thorsteinsson, 'The census of Iceland in 1703', Nordic Statistical Journal, 8 (1929), 362-70; H. Palmström, 'The census of population in Norway, August 15th 1769', Nordic Statistical Journal, 8 (1929), 371-80; H. Gille, 'The demographic history of the northern European countries in the eighteenth century', Population Studies, 3 (1949-50), 3-18.
3. K. H. Connell, The Population of Ireland, 1750-1845 (Oxford, 1950), p. 3.
4. For details of these see Kaare Ofstad, 'Population statistics and population registration in Norway: Part 3, Population censuses', Population Studies, 3 (1949-50), 66-75.
5. A copy of the census schedule together with the instructions to the enumerators is in Wessel Berg, Kongelige rescripter, resolutioner og collegial-breve for Norge i tidsrummet 1660-1813, IV, 1797-1813 (Christiania, 1845), pp. 273-80.
7. Census of Norway 1801. My analysis of enumerators' returns for parishes of Nes and Ål in Hallingdal. These are now in the Riksarkiv, Oslo. For liability for military service see T. R. Malthus, An essay on population (Everyman edition, London, 1914), book 2, p. 155.
8. This led to a serious uprising in and around Bergen in 1765. Karen Larsen, A history of Norway (Princeton, 1948), pp. 322-3, About 4000 regular soldiers were excluded from the census. For a further discussion of this census see notes in statistical appendix.
9. The schedule and the instructions used in the rural districts appear in Den norske rigstidende for 1815, no. 9. The enumerators were asked to place the population in eight-year age groups. A footnote to the schedule - it looks very much like an afterthought - asked that a separate note be made of the number of male and female children under four years of age. An examination of the original returns (now in the Riksarkiv, Oslo) reveals that in at least the following parishes the under four-year groups was not included in the under eight-year group: Røyken (Drammens deanery), Spydeberg (Øvre Borgesyssel); Gjerdrum (Øvre Romerike); Botne and Ramnes (Jarlsberg); Solum (Bamble); Manger and Hosanger (Nordhordland); Ona (Nordmøre); Hadsel (Vesterålen); Loppa and Alta-Talvik (Vest-Finnmark). In at least two of these parishes this under four-year group was not included in the population total for the parish as a whole. This made the total for the parish of Spydeberg, 1656 instead of 1822 and for Gjerdrum 1186 instead of 1297.
10. Norges officielle statistik, ottende række, Tabeller over folkemengden i Norge den 31te December 1845 samt over de i tidsrummet 1836-1845 ægteviede, fødte og døde (Christiania, 1847).
11. Norges officielle statistik, ottende række, Tabeller over folkemengden i Norge den 31te December 1845 samt over de i tidsrummet 1836-1845 ægteviede, fødte og døde (Christiania, 1847).
12. O. Vig, 'Nogle ord um folketælling m.m.', Folkevennen, 4 (Kristiania, 1855), 306 notes that farmers in 1835 thought the agricultural part of the census was a prelude to new taxation. These fears proved groundless as in 1836 the land tax was abolished. For this reason the agricultural census of 1845 was more complete.
13. Eilert Sundt, Om dødeligheden i Norge (Christiania, 1855), p. 23.
14. Norges officielle statistik. Ældre række, C. no. 1, Tabeller vedkommende folkemængdens bevægelse i aarene, 1856-1865 (Christiania, 1868-9), p. vii.
15. Gunnar Jahn, 'Folketellingene 1801 og 1815 og befolknings-forholdene dengang', Statsøkonomisk Tidsskrift, 43 (Oslo, 1929), 202,