Labouring Poor in Early Modern Sweden?

We know very little about the unlanded population (obesuttna) in Sweden before 1750. Still, knowledge of this group is vital not only to our understanding of the proletarianization process, but also to the understanding of rural Sweden before the changes of the 18th and 19th centuries. This article takes a step towards filling this gap. Tax registers and church records from 17th-century Västmanland shows substantial proportions of unlanded people, i.e. crofters, cottagers, soldiers, artisans, and lodgers, around 1640 and 1690. The age structure of this group, which can be studied for the latter part of the period, shows that it mostly consisted of people in working age. In conclusion, the article argues that there was a significant group of smallholders and wage labourers in 17th-century Sweden, comparable to the labouring poor of other European countries.


Introduction
In Swedish and European historiography, the part of the rural population with little or no land is generally associated with the proletarianization process, the agrarian revolution, and the rise of capitalism, i.e. with the transition from one economic regime to another. Their growth in number over time has been used as an indication of a new mode of production, characterized by wage labour and the dissolution of the household as the unit of production. 1 Different classification practices in historical sources as well as among historians makes comparisons difficult. Still, the fact remains that a large number of studies have shown that a substantial part of the population in medieval and early modern Europe did not have enough land to support themselves. Tilly estimated, for example, that the proportion of 'proletarians' in Europe increased from 30% in 1500 to 67% in 1800. 2 In late 17th-century France, agricultural labourers could make up as much as 90% of a village's population. 3 In Brandenburg, smallholders accounted for 40% of the rural population in the early 1600s. 4 According to Dyer, more than half of all households in rural England lived on holdings inadequate for the support of a family as early as around 1300, while Hindle reports that the proportion of 'labourers' in the English rural population increased from 20-30% in 1520 to 50% in 1650. 5 These results have generated a greater interest in the group of so-called labouring poor, i.e. people who neither had enough land nor were paupers, but depended on wage work for their survival. 6 Increasingly, this group has attracted attention not only as carrier of change but as vital for our understanding of the old regime as well, including patterns of production, consumption, and credit networks in early modern Europe. 7 In Sweden, labourers and smallholders are generally considered to have been of little importance before the 18th century. Research and overviews focus instead on the landed peasants (bönder), i.e. freeholders and tenants. 8 When landless or semi-landless groups are studied, it is most often in relation to the transition problemthe agrarian revolution and the development of capitalismand to the increasing number of unlanded people (obesuttna)crofters (torpare), cottagers (backstugusittare), and lodgers (inhyses)that has been observed from the middle of the 18th century (see Table 1). 9 Our knowledge of this group before the 18th century is much more limited. This paper takes a step towards filling this gap.
This article investigates the size and the composition of landless or semi-landless groups in the 17th century. How large was the unlanded population? How did this vary over time and space? Who were they? The approach is source pluralistic. 10 There are no general registers of crofts and cottages in the realm or any survey of the people who inhabited these small houses or the way they made a living. A variety of sources must be consulted and compared, therefore. This makes a limited geographical scope necessary. For reasons given below, the study is based on empirical data from 17thcentury Västmanland in central Sweden.
After discussing previous literature and introducing the study area and the vocabulary, the next section uses livestock registers to map the unlanded population in Västmanland in 1640. It demonstrates a substantial group of landless or semilandless households, but also great variation. This pattern is strengthened by a comparison with other regions as well as to other sources from Västmanland. The third section discusses the development from 1640 to the mid-18th century, when population statistics are available. Poll tax registers and church records from different Västmanland parishes suggest that the unlanded population in the study area increased in the late 17th century and decreased again after 1700. The fourth section look more closely at the characteristics of this group in the late 17th century. It shows that this group consisted mostly of people in their working age. Together, the results point to the importance of including labourers and smallholders in the analysis of Swedish rural society before the transformations of the 18th and 19th centuries.

A Swedish exception?
According to Gadd, landless people and smallholders made up a smaller share of the population in early modern Sweden than in, for example, France. 11 This may very well be true. But if that is the case, we need to ask why. One possible explanation would be that the Swedish agrarian economy had little need for rural labourers. Kula has argued that it was the demands of the demesne that motivated the division between peasant households with adequate landholdings and households with too little land for their own subsistence. More substantial peasants, who could afford ploughs and plough animals, used these to cultivate the arable land of the demesne, whereas smallholders did whatever work that required simpler tools or their bare hands, such as ditching and harvest work. 12 For Sweden, Torbrand's analysis of the work organization at the estate of Johannishus in around 1730 demonstrates a similar division of labour between landed peasants (bönder), who did most of the fieldwork, and crofters (torpare), who mainly performed various kinds of services for the estate, such as handicraft, building, and forest work. At peak seasons and during large construction projects, temporarily employed labourers supplemented the workforce. 13 There were relatively few large demesnes in Sweden. Kula's approach may be less relevant, therefore. On the other hand, Sweden had many ironworks and comparable industries. On such units, a corresponding division of labour has been found. Landed peasants, who had draught animals, did most of the transportation work, whereas crofters produced charcoal and did manual work when needed. 14 Thus, Kula's argument may be relevant after all.
Another approach to the existence of labouring poor, put forward by Hilton, stresses the seasonal character of agriculture. The need for labour varied over the year and the poor formed a flexible labour reserve that could be utilized at peak periods, i.e. haymaking and cereal harvest. The result was a reciprocal relationship between rich and poor, in which the former relied on the work of the latter, whereas labourers needed the wages paid by the landholders. 15 Hilton's prime interest was the conditions of the English peasantry, but studies of other parts of Europe have demonstrated similar reciprocal relationships. 16 Unlike Kula's argument, this model does not presuppose large estates, but is equally applicable to smaller peasant holdings.
Then again, seasonality is mainly a characteristic of arable farming, whereas mixed farming dominated Sweden. The raising of livestock supplemented, and in some areas was more important than, grain production. Following Mitterauer, cattle-raising societies relied on live-in servants, who were employed all year round, rather than temporary day labourers, because taking care of livestock was continuous and required a permanent work force. 17 On the other hand, temporary labour was seldom completely absent, and haymakingfundamental in Sweden, where long stalling periods were necessaryremained a seasonal peak. 18 Both of the above approaches stress the economic function of a group of labouring poor in the agrarian system. The main argument is, then, that such a group had only a limited function in the Swedish system. In 17th-century Norway, however, which shared many of the agricultural features with Sweden, many husmenn typically worked as temporary labourers on peasant farms. 19 In Norway, too, the traditional view has been that the husmenn were a relatively insignificant group before the increase during the course of the 18th century. 20 Yet, as remarked by Naess, the fact that husmenn were treated as a separate group in taxation registers from as early as around 1600 speaks to the point that they were already significant then. 21 Recent studies have found that a system of Norwegian husmenn was established in the middle of the 17th century. 22 One important aspect in which Sweden differed from Norway, and which constitutes another possible explanation for an assumed lack of landless households in early modern Sweden, is law. Neither labour laws nor regulations on the establishment of cottagers and crofts seem to have allowed much room for rural labourers. In 1540, Gustav Vasa forbade day labour in rural areas. 23 The statute of servants of 1664 and later, similar ordinances declared that people who lacked a homestead, wealth, or a trade were compelled to be employed full-year as servants. Lodgers should not be tolerated. 24 In order to protect the forests in the interest of the mining industry, crofts and cottagers were not allowed on commons. 25 Crofts that were built on freehold land and paid rent to neither crown nor the nobility were to be taxed or torn down. 26 To ensure the rent-paying ability of the farm, the right to farm subdivision was restricted. 27 Both Elgeskog and Wohlin assert that these regulations were observed only to a limited degree. 28 Still, the restrictive laws are held as arguments for limited numbers of crofters and cottagers in Sweden before 1750. While pointing out the lack of research on the subject, Rantanen claims that, as Swedish peasants were forbidden to set up crofts on their land, they hadunlike their Norwegian counterpartsin principle no access to crofters' labour before 1750, when restrictions were lifted. 29 Svensson reaches the same conclusion but argues somewhat differently. According to his argument, peasants were allowed to set up crofts on their land before the mid-18th century, but as this yielded higher taxation, it 'naturally held back the establishment of crofts on peasant land'. 30 Just like the arguments based on the agrarian system, these assumptions are theoretical. They need to be tested empirically.
When actual numbers are studied, they suggest non-negligible numbers of people with little or no land in 17th-century Sweden. As Wohlin pointed out, the large number of crofts in 1750 indicates that many crofts had been established earlier, despite regulations. 31 In a study of an early church register, Friberg and Friberg identified 21 out of 65 households, or one-third, in the parish of Tillberga, Västmanland, in 1632 as 'lodger households' (inhyseshushåll). They even assumed that this group was under-registered. 32 In the Finnish part of the Swedish realm, there seems to have been a considerable group of landless people. In some rural areas, one-fifth or one-quarter of the population has been classified as landless in the 1630s, and in the 1690s, as many as 50% had no or very little land. 33 Thus, there are reasons to look more carefully into the group of crofters, cottagers, and lodgers in Sweden and ask to what extent they existed before 1750. Studying this group has a bearing on our understanding of Swedish rural communities in the early modern period more generally. The sheer existence of a part of the population with no or too little land for their own survival negates traditional conceptions of a pre-modern rural world that consisted of self-sufficient and discrete units of production and consumption. Instead, it points to a more heterogeneous character of the rural community and raises questions about the prevalence and structure of the labour market before the agricultural revolution.

Why Västmanland?
The data used in this article are from 17th-century Västmanland, a province in central Sweden, just north of Lake Mälaren. There are two main reasons for this choice. First, when compared to other Swedish areas, Västmanland has an unusual amount of goodquality sources from the 17th century due to its early parish registration. Consequently, Västmanland has earlier been the subject of many historical demographic studies. 34 Second, Västmanland had a diverse economy, with forests and mining areas in the northern part and large arable fields in the south. In some areas ironworks dominated; in others, large estates or small peasant farms characterized the local economy. Thus, within Västmanland, the conditions for a group of labouring poor are likely to have been very varied. This allows for a nuanced analysis of this group.
Special attention will be paid to the juridical district (hundred or härad) of Snevringe and the parish of Björskog. Again, this choice rests upon the availability of suitable sources, especially parish registers, and varying local economies. Snevringe, which comprised nine parishes, included both grain-producing areas by Lake Mälaren and wooden areas dominated by ironworks at the border to Bergslagen. There were large manors in the region, but also peasant-dominated areas. 35 Grain production, small family farms, and a few larger estates characterized Björskog, in the arable region by Lake Mälaren. 36 No claim is made that Västmanland would be representative of Sweden as a whole. According to the figures presented by Wohlin, the share of obesuttna in Västmanland was larger than the national average both in 1751 and in 1850 ( Table 1). The same may very well be true of earlier centuries. It is likely that the situation differed from region to region. This article will only briefly address to what extent the patterns found here apply to other parts of Sweden, as well.

Labouring poor = obesuttna?
'Labouring poor' is a term derived from an English context. It cannot, without discussion, be equated with the Swedish term obesuttna, which is a general designator for rural inhabitants who did not live on a cadastral (i.e. mantalssatt) farm. 37 While the English term stresses the group's dependence on their labour, the Swedish word stresses their lack of land. On the other hand, the lack of land is exactly what made them dependent on their labour. They were, in other words, proletarians, or rather semi-proletarians. Few had no access to land at all. Wage work in early modern Europe often included different kinds of entitlements. 38 The difference between landed peasants (bönder) and obesuttna is formal. Ideally, and in the eyes of the authorities of the times, this distinction should coincide with a real difference. The extent to which it did is an empirical question. Winberg has argued that, although there were exceptions, the formal and the real boundary between landed and unlanded corresponded roughly. 39 Other research has demonstrated that there were crofts that were of the same size as a small peasant farm in terms of arable acreage, meadows, and livestock. 40 Acreage data from the 17th century exist for only a few crofts in the study area. A map of the manor Almö (just outside Snevringe) from 1691 shows two smallholdings with 0.2 and 0.6 hectares of arable land, respectively. 41 A croft on a peasant farm situated in the parish of Säby (in Snevringe), mapped in 1688, had a somewhat larger holding of 0.86 hectares. In comparison, the three peasant farms in the same hamlet had, on average, 4.7 hectares of arable land. 42 In the hamlet of Gräsnäs in Björskog, a croft had 0.5 hectares of arable in 1690, compared to an average of 9.7 hectares among the nine peasant farms in the same hamlet. 43 A preparatory map of an area in the parish of Ramnäs (in northern Snevringe) made in 1676, lists 12 crofts. The average acreage was 1.6 hectares of arable land and the size of fields varied from 0.3 to 3.3 hectares. 44 As crofts were generally established on less fertile soil, it is likely that they yielded less than peasant farms. Moreover, crofters did not have access to common resources, such as woods, in the same way as landed peasants did. But even if there were no sharp distinction between a better-off crofter and a landed peasant with a small farm, in economic terms, any calculations of the proportion of unlanded people (or labouring poor) in a given area depends on the classifications made in the sources, which more likely refer to formal delimitations than real ones. In this study, numbers of formally landless are used as a proxy for numbers of real landless or semi-landless, therefore.
The group of obesuttna includes crofters (torpare), cottagers (backstugusittare), lodgers (inhyses), and soldiers and artisans without taxed land. The distinction between a torp (croft) and a backstuga (cottage) was often imprecise, in terms of both acreage and conditions of possession. While crofters were supposed to have larger holdings and more secure possession, this was not necessary the case and it cannot be taken for granted. Moreover, the words backstugusittare (cottager), inhyses (lodger), and husfolk (house-people) were sometimes used interchangeably. 45 A person who was referred to as an artisan in one source may be called crofter in another. 46 Live-in servants are sometimes counted among the unlanded and the labouring poor. 47 Naturally, this increases the proportion of the landless population in a given area and is one example of how classification practices of historians affect the result. Servants were indeed often landless and made a living by wage labour; yet they belonged to the household of their master. 48 In this study, they are not included among the unlanded. The existence of servants, who were employed for the whole year and lived with their masters, is part and parcel of our common understanding of the landed peasant household as an economic unit in early modern Sweden. People living apart from these landed households indicate the existence, and the importance, of casual labour and/or other forms of exchange relationships between households.

Crofters and lodgers around 1640
Tax registers and church records are the chief demographic sources of Swedish areas before the introduction of national population statistics (Tabellverket) in the middle of the 18th century. Tax registers exist in large quantities, but as they were created with the purpose of collecting taxes rather than counting people, they often left out poorer parts of the population. 49 According to Juttikala, the fact that poll tax registers (mantalslängder) from the 17th century include very few households that did not have a farm does not mean that such households did not exist. Instead, it is due to the incompleteness of the source. 50 There is one kind of tax register from the 17th century that is likely to include landless and semi-landless households to a greater extent, however. Between 1620 and 1641, a tax on possession of livestock was levied as a complement to the ordinary taxes on land. 51 As a basis for the collection of the tax, listings of each household's possession of horses, oxen, cows, sheep, etc. were compiled for every parish. The tax was supposed to include not only landed peasants, who paid annual rents on the basis on the alleged capacity of their farm, but also crofters and lodgers: So shall also others who live in the countryside, although they do not yet pay annual rents, but have parts by the peasant [hafwa Deel inne medh bondhen], or else serve, or else live as lodgers, and support themselves by work, no less than the peasant himself be obligated to pay the said contribution for their livestock. 52 As the quote makes clear, the Swedish authorities acknowledged a group of rural inhabitants who did not have a farm, but supported themselves by work, and assumed that they had livestock. Given what we know about labouring poor in other parts of Europe, this assumption is plausible. The keeping of livestock was an important part of the ways in which this group made ends meet. 53 The livestock registers offer unique opportunities to systematically study the proportion of unlanded households over large areas. Yet, it must be acknowledged that they generally list only households with livestock. Some registers include a group of people who were designated as 'poor' (fattiga or utfattiga). They were described as old or disabled and were not taxed for having any livestock. This group seems unsystematically recordedonly registers from five out of 65 parishes include such a listand has been left out in the following. (In total, this group make up 0.5% of those listed.) The proportion of unlanded households presented in Figure 1 should therefore be seen as minimum shares, i.e. only those households that had (taxed) livestock.
Apart from the category 'poor', some registers include headings like 'crofters' (torpare), 'lodgers' (inhyses), or 'house-folk' (husfolk), but not all registers have headings of this kind. To identify the unlanded population, I have used another criterion. Because a part of the tax was based on the household's area of arable land, the registers note not only the possession of livestock but also arable land under cultivation as well. The reported numbers were fiscal measurements rather than real acreage. 54 Hence, we should think about those without any registered land as formally, or fiscally, landless, not necessarily actual landless. This does not make them different from the obesuttna of 18th-century statistics, however, and we can assume that if they had land, the amount was limited. To a large degree, the group corresponds to to those listed under headings like 'crofters' and 'lodgers' in registers in which such headings exist, although the overlap is not absolute. Figure 1 maps the proportion of listed households in the 1640 livestock registers without any registered arable land in 65 parishes in the county of Västmanland. For simplicity, I regard every name listed in the registers as head of a household. 55 The registers list 5,148 households in total, excluding people listed as 'poor'. Overall, 26% of these households lacked sown land according to the registers. Hence, according to the livestock registers, every fourth household in Västmanland was fiscally landless in 1640.
It is beyond the scope of this paper to investigate whether Västmanland was exceptional, but a comparison to two other regions might shed some light on the question. The county of Uppsala, situated close to Västmanland and with a similar regional economy, displays a similar pattern. In the county as a whole, landless households that paid tax for livestock made up 25%, i.e. almost the same percentage as in Västmanland. In Uppsala, too, the proportion of landless households in a given parish differed markedly, from 6 to 49%. 56 The county of Jönköping, on the other hand, was different. In 10 out of 99 parishes in Jönköping, there were no households without registered land at all. Unlanded households made up more than 40% in one parish only and, overall, no more than 10% of the households in the county were landless, according to the livestock registers. 57 Thus, Västmanland was neither unique nor the same as other Swedish regions. This is hardly surprising. More notable are the differences between parishes within the same region. Obviously, local socioeconomic factors were a determinant, but in order to explain these variations in detail, a much more full analysis is needed.
Here, instead, the proportion of fiscally landless in the livestock registers will be contrasted with information given in church records. Unlike tax authorities, the church was a source-producing body that was more interested in people than in the carrying capacity of their property. In contrast to tax registers, however, church records from the first half of the 17th century exist for only a few areas. Most of them belong to the diocese of Västerås, which includes Västmanland.
Contemporary church records suggest an even larger proportion of unlanded households than the livestock registers. As referred to earlier, Friberg and Friberg found that lodgers made up one-third of the households of Tillberga in 1632. 58 According to the livestock register, the proportion of households without registered land in the same parish in 1640 was 14%. Perhaps there had been a real decrease, although under-registration is a more likely explanation. The fact that the livestock register of Tillberga 1633 only includes four landless households (and 40 landed) reinforces this. 59 The most complete church register from the middle of the 17th century is the catechetical examination register (husförhörslängd) from the parish of Björskog, 1643. 60 One-third of the 125 households in the register did not live on a farm but in crofts (torp) or on the 'land of the hamlet' (byns ägor). 61 Whether this latter group should be considered crofters, cottagers, or lodgers is impossible to say, and perhaps it made little difference to them. In comparison, the livestock register of 1640 lists 112 households in Björskog, of which 20 households fell under the heading 'crofters and house-folk', eight soldiers and one parish clerk lacked arable land. Together, this adds up to 26%. Hence, the church registers show that the livestock registers underregistered the landless population in Björskog, too, which is unsurprising considering it listed only those who had livestock.
On the other hand, the church registers make clear that the landless households were smaller than the households of landed peasants. This has been demonstrated earlier for other areas and for the 18th and 19th centuries. 62 If we count individuals instead of households, the people who lived in unlanded households made up 18% of the population in both Björskog and Tillberga. 63 There are no contemporary catechetical examination registers from the area of Snevringe. However, in the parish of Berg, the volume that holds the oldest church accounts includes a 'parish register' (sockenlängd) from around 1640. It lists the hamlets in the parish, the farms and their possessors (i.e. landed peasants), but also crofters, soldiers, and artisans. The number of peasants with a cadastral farm roughly coincides with the number in the livestock register (26 and 23, respectively); however, whereas the livestock register of 1640 lists only nine husfolk ('house-folk'), the parish register includes 23 crofters without a farm plus 24 soldiers and six craftsmen who were not included among the peasants or the crofters. 64 In a similar register from the parish of Stora Rytterne, dated 1641, crofters, lodgers, and soldiers make up 58% of those listed. 65 The registers do not make it clear whether these people should be regarded as heads of their own households or if they were lodgers, but they support the conclusion that there was a significant group of landless or semi-landless in Västmanland in the 1640s.
The development in the following century Table 2 relates the proportion of fiscally landless people, according to the livestock register of 1640, to the proportion of unlanded male heads of households in 1751, according to Tabellverket, in Björskog and parishes in Snevringe with available data. Two ways of calculating the proportion of unlanded according to Tabellverket has been used. The left column presents proportions in accordance with the method used by Nils Wohlin, which relates the numbers of crofters, cottagers, and able-bodied lodgers in relation to the number of landed peasants. 66 The right column shows the numbers of unlanded defined more broadly, including soldiers, craftsmen, and disabled people, as proportions of all male heads in the parish. The exercise illustrates the impact of definitions for calculating the landless and semi-landless population. For example, in Ramnäs and Sura, all male heads under the heading 'people at the ironworks' (Bruksfolk vid Bergverken) have been counted as unlanded. This is certainly doubtful when it comes to master smiths and apprentices, but smith's assistants, charcoal workers, and day labourers had very much in common with other crofters. 67 In Tabellverket, these groups are lumped together, which makes it impossible to separate them without consulting other sources.
Remember that Table 2 presents only male heads, as a proxy for households, in order to make the numbers comparable to the proportions presented by Wohlin (see Table 1). As will be demonstrated later, and as has been shown by earlier research, women made up the majority of the unlanded population. Thus, had the numbers of Table 2 included women as well, the proportions of unlanded people in 1751 would have been higher.
All of this makes clear that the columns of Table 2 are not really comparable. Still, they do suggest an increase in the proportion of unlanded households in the study area between 1640 and 1751. There is no reason to assume a priori that the growth had been linear, however. In fact, available data point to a substantial increase in the latter part of the 17th century, followed by a decrease after 1700. Figure 2 shows the share of unlanded households in three parishes in Snevringe, according to poll tax registers, every 12th year from 1640 (when it is based on livestock registers) to 1748. In addition, it shows the size of the taxed population in these parishes, taken together. To construct the figure, every entry in the registers has simply been counted as one household. Using cadastral registers as a complementary source, I have identified households that did not live on taxed farms (and/or were not persons of standing) and defined them as unlanded. In addition, early registers include separate listings of unlanded households under headings such as 'Crofters and lodgers'.
The socioeconomic settings of the three parishes differ. Lilla Rytterne, by Lake Mälaren, was dominated by great noble estates; Munktorpthe largest by far of the threehad some large estates, but was otherwise characterized by small peasant farms; Svedvi, on the borderland between the fertile plains of Lake Mälaren and the woodlands of Bergslagen, was dominated by peasants and an ironworks. These differences probably explain some of the variance between the parishes.
When it comes to tax registers, variation over time is otherwise likely to be the result of changing registration practices as well as real change. For example, the great fluctuations in the number of unlanded households in Lilla Rytterne before 1688 seem more due to changing policies for registering units on tax-exempted nobility land than to real change. Also, the decrease in taxed population in the first decade after 1640 is partly explained by new registration practices after the abolition of the livestock tax. Between 1641 and 1643, i.e. in only two years, the taxed population of the entire hundred of Snevringe decreased by 8.5%. Not surprisingly, it was mostly the unlanded population who had disappeared from the registers. Friberg and Friberg have demonstrated under-registration of crofters and lodgers in poll tax registers in Västmanland in the first half of the 17th century. 68 Correspondingly, and as evident from Figure 2, when the taxed population increased in the late 18th century, the inclusion of unlanded households explains most of this.
Still, a real growth in the proportion of landless and semi-landless households in the late 17th century is likely, given what we know about the area and time period in question. Although estimates on population growth rate vary, historians agree that the Swedish population increased during the 17th century. 69 During the same period, agricultural production, at least the part that was captured by tax records, seem to have stagnated. 70 This is true also for Västmanland. 71 In this relatively densely populated part of Sweden, there was probably not much arable land left uncultivated for the establishment of new farms. Population growth created an excess of people without land.
At the same time, the production of iron and other metalsvital to the economy of the study areawitnessed a steady upsurge in the 17th century. 72 The mining industry provided an income to many people, landed peasants as well as unlanded. The expansion of noble estates and, later, the introduction of the military allotment system (indelningsverket), in which hired soldiers were given small crofts, also created new opportunities for those of the population who did not have access to a peasant farm.
The decrease after 1700, both in population and in the proportion of unlanded households, is also likely to have been real. Bad harvests in the latter part of the 1690s had severe demographic consequences. 73 Friberg and Friberg found that the population of Tillberga decreased by 25% during this decade. 74 According to Jutikkala, the extreme mortality rates in the 1690s and the following Great Northern War meant that the unlanded population in Finland, which had been significant and increasing during the 17th century, was severely diminished: they either died or took over abandoned farms and became landed peasants. 75 In the parishes of Munktorp and Svedvi, the proportion of unlanded households were very much the same from 1688: about 40% at the end of the 17th century and about a third in the following decades up until 1748. Likewise, in Berg and Säby, two neighbouring parishes, both the proportion and the development of unlanded households were much the same. Thus, the large local variations that seem to have characterized the 1640s had disappeared. In the manor-dominated parish of Lilla Rytterne, however, the share of landless households was larger and exceeded 50% from 1688 onwards.
Available church records support the idea of an increasing number of unlanded households in the latter part of the 17th century. Table 3 shows the proportion of Sources: ULA, Bergs kyrkoarkiv A I:1; ULA, Björskogs kyrkoarkiv AI:2; ULA, Sura kyrkoarkiv AI:1.

LABOURING POOR IN EARLY MODERN SWEDEN
households and individuals who have been identified as crofters and lodgers in three parishes in Västmanland, where catechetical examination registers exist. The lists are not complete; most significantly, they do not include small children. In the case of Björskog, I have added children with the help of baptism and burial registers, but the figures do not include small children in the other two cases. Table 3, too, points to an increase of unlanded households in the areas under study compared to the 1640s. Between half and two-thirds of all households belonged to this category in the 1680s. While the number of landed peasant households in Björskog increased by 19% between 1643 and 1688 (due to subdivision of farms), the number of unlanded households more than doubled during this period and the population living in these households increased by 95%. 76 Björskog was characterized by arable farming and few large manors. Here, 30% of the population lived on crofts or as lodgers in 1688. In Sura, this proportion was much higher (60%). A large number of crofts related to the ironworks of the area probably explain this.

Crofters and lodgers around 1690
The previous section demonstrated the existence of a significant and increasing number of landless and semi-landless people in 17th-century Västmanland. However, for this group to qualify as 'labouring poor', we need to know more about who these people were. Were crofters and cottagers before the 19th century, as has been claimed, only 'a small group of outcasts, halt and lame persons'? 77 Their sheer number speaks against this, but this does not necessarily mean that they constituted a rural proletarian class of their own. Earlier research has assumed that, before the agricultural revolution, being a cottager or a lodger was typically a stage in the life cycle. The landless were either young, waiting to acquire a farm of their own, or old, retired peasants. 78 For Norway, Dyrvik found support for this idea in the age distribution of husmenn in 1664-1666, whereas Kongsrud, for the same point in time but another area, did not. 79 Thus, to what extent did the group of landless and semi-landless households consist of able-bodied people in working age? Among the 480 households that were identified as unlanded in the poll tax registers of Snevrige in 1688, one-fifth included one or more persons who were described as old or disabled. 80 Still, the great majority was not described in these terms. To dwell deeper into this question, we need age data. Such data are rare for the 17th century, but exist in some parish registers. Among those are the catechetical examination register from Björskog in 1688. Children under the age of 10 are unevenly registered, but age data exist for 92% of the remaining part of the listed inhabitants. Figure 3 shows the age and sex distribution of the population, 10 years and above, who resided in unlanded and landed peasant households in Björskog in 1688 (the households of persons of standing, representing 10% of the total parish population, are not included). It shows, first, an uneven distribution of the sexes. While women made up 52% of the landed peasant population in Björskog, they constituted 62% of the unlanded population. Women were especially prominent among lodgers. Of 38 lodgers in Björskog in 1688, 30 were women. This pattern was not unique to Björskog. The catechetical examination registers from Berg and Sura parishes (both in Snevringe) display similar sex ratios. 81 Over-representation of women among the unlanded population is also in accordance with what earlier research has found for later centuries. 82 In the 17th century, there was a large female surplus in Sweden due to the many wars of the period. 83 In the study area, this surplus did not characterize the landed households, but was mainly found among crofters and lodgers.
Unlike the landed peasants of the parish, the age distribution of the landless and semi-landless households in Figure 3 do not exhibit the classical pyramid shape of a pre-industrial population, with diminishing proportions of the total population for every age group. Instead, all age groups between the age of 10 and 59 are of a similar size. This, too, is in accordance with the results of studies of crofter households in the 18th and 19th centuries. Here, too, both men and women were under-represented in younger age groups. The explanation is that sons and daughters of unlanded families left home earlier and to a greater extent, whereas landed peasants attracted people in exactly these age groups as servants. 84 In all three parishes listed in Table 3 -Björskog, Berg, and Surafarmhands and maids who lived on peasant farms or manors made up between 17 and 21% of the registered population. Thus, a significant servant population existed alongside the large group of landless and semi-landless households in the area. As Lambrecht, among others, has argued, the existence of servants should partly be explained in the light of the smallholding economy. Apart from the fact that sending out children to serve meant fewer mouths to feed, part of the wages of young farmhands and maids were sometimes paid directly to the parental household. 85 The fact that there were relatively few people in their twenties living in landless and semi-landless households speaks against the assumption that being a crofter or lodger was a phase before the acquiring of a farm. Instead, the unlanded population was somewhat older than the peasants were. The unlanded population made up 40% of the total population between 30-59 years old, and 48% of those above 60 years old. Probably, some of them were retired peasants.
Still, landless and semi-landless households existed in all age groups. This is true for different sub-groups, as well. Table 4 shows that all age groups are represented among lodgers as well as in the households of soldiers and crofters. On average, soldiers' households were younger than the households of crofters, which, in turn, were younger than the lodgers. This speaks to differences between groups, but also to significant variations within each group.
By itself, age says little about the ability to work. A 20-year-old may be disabled or severely ill. But it is highly unlikely that all these people were paupers. For some, being a lodger or a crofter might have been a life-cycle phenomenon, but the age distribution does not support that this applied to the entire group.
The life cycle hypothesis does not find support in the burial registers of Kolbäck and Säby parishes, either. These registers include obituaries (personalier), short descriptions of the life and conditions of the deceased, that allow us to relate ages of death to current social position, as well as to previous experiences. In total, the burial registers of Säby and Kolbäck list 674 burials between 1687 and 1699. Of them, 304, or 45%, are classified as landless or semi-landless people. In most cases, my classification is based on place of living, i.e a croft or 'on the lands of the hamlet' (på byns ägor), but also, when existent, on titles. Information on age of the deceased is given in 91% of all cases and 90% of the landless and semi-landless cases. Figure 4 shows the ages of unlanded people, as well as landed peasants, who were buried in Kolbäck and Säby in 1687-1699. Ages at death do not necessarily reflect the age distribution among the living, but the chart clearly demonstrates that there were landless and semi-landless people in all age groups. What is more, their pattern is very similar to that of the landed peasants, exceptagainfor people in their teens or twenties (who probably had left home to works as servants). The result does not support that being a crofter, cottager, or lodger represents certain phases in life. Among men (discounting children under the ages of 15), the difference in mean age is negligible: 48.1 for landed peasants and 48.5 for Obituaries were written very irregularly, leaving imperfect data that are difficult to quantify. 86 It may be worth noticing, however, that of about 100 landless or semilanded people over the age of 40, only six were said to have previously possessed a farm of their own.
To sum up, as far as the studied data from Västmanland in the late 17th century goes, the majority of the unlanded population were working age and not evidently disabled. Though it might have been true of some, being landless or semi-landless was not typically a life-cycle stage. Thus, it seems that there were a significant number of able-bodied people in 17thcentury rural Västmanland without enough land for their subsistence, comparable to the groups of labouring poor in other European countries.

Conclusion
This article has investigated the size and the composition of landless or semilandless groups in 17th-century Sweden. Depending on time, place, and what source is used, shares differ. Proportions of unlanded households in the livestock registers from Västmanland in 1640 varied from under 10% to more than 50%. The average, one-quarter of all listed households, was the same as in the county of Uppsala, but significantly higher than in the county of Jönköping. The proportion of unlanded households was larger in church records than in tax records, which is in agreement with the assumption that tax records under-register poor households. While the livestock registers include many households without taxed land, they omit households with no animals. Proportions of unlanded households in tax records must be seen as minimum numbers, therefore. Church records, which capture a larger part of population, exist to a much more limited degree, however.
The presented evidence indicates an increase in the unlanded population of the study area and a decrease after 1700. Regional differences were great and Västmanland was hardly typical. Still, the results of this study suggest that the history of the unlanded population in Sweden should be revisited and revised. It is a history that started long before 1750. It is not only a history of the transition to another economic regime, but also of continuity and variations within an older system. Over this longer period of time, it is not only a history of linear growth (as Table 1 suggests), but of large fluctuations (as in Figure 5). Restrictions aside, crofts and cottages were built not only around manors and ironworks, but also on freehold land in peasant-dominated areas. Limited numbers of crofts and lodgers cannot be assumed from regulations only.
Has the article confirmed the existence of a significant group of labouring poor in early modern Sweden? What has been demonstrated is a substantial group of smallholders and lodgers in the 17th century. To a significant degree, they were of working age. Their economic significance, in different socioeconomic contexts, needs to be further explored. The high proportions of live-in servants in the study area shows that the labour of crofters and lodgers was not a substitution for servant labour but existed alongside it. Still, the presence of these groups indicates the importance of casual labour and/or other forms of exchange relationships between different types of households in early modern Sweden. Recent research has emphasized the many ways in which households were entangled in networks across household borders as well as the prevalence of multi employment in the ways that most people made their living. 87 This opens up a view of rural Sweden before the agricultural revolution that differs from the old conception of a peasant economy consisting of self-sufficient and discrete units of production and consumption. In order to learn more about this economy, we need to know more about how the unlanded population made a living and how they interacted with landed estates, ironworks, and the landed peasantry. This would affect the history of proletarianization in Sweden as well as our understanding of rural society in early modern Sweden more generally.

Disclosure statement
No potential conflict of interest was reported by the author.