Land distribution and inequality in a black settler colony: the case of Sierra Leone, 1792–1831

Land distribution is considered to be one of the main contributors to inequality in pre-industrial societies. This article contributes to the debate on the origins of economic inequality in pre-industrial African societies by studying land inequality at a particularly early stage of African economic history. The research examines land distribution and inequality in land ownership among settlers in the Colony of Sierra Leone for three benchmark years over the first 40 years of its existence. The findings show that land inequality was low at the founding of the Colony but increased substantially over time. We suggest that this increase was enabled by a shift in the type of egalitarianism pursued by the colonial authorities, which was reflected in a change in the redistributive policy applied, which allowed later settlers to appropriate land more freely than had been previously possible.

S ierra Leone has been considered the epitome of a failing African state, with violence and poverty spiralling out of control on several occasions. 1 Struggles over resource distribution have fuelled conflicts between groups ever since the country was unified in 1896, and eventually contributed to the outbreak of a civil war there in the 1990s. 2 Yet, earlier in its history, a pillar of the foundation of the Colony of Sierra Leone was a striving towards equitable resource distribution. 3 A growing number of studies on historical inequality argue that the roots of inequality in a country can be traced back to its early stages of development. 4 Colonialism is one element that scholars have identified as influencing inequality. In a series of seminal works, Acemoglu et al. have claimed that the colonial institutional framework tends to persist long after independence, effectively influencing resource distribution and development in the long run. 5 Scholars disagree nonetheless on why different types of institutions (that is, those more or less geared towards an (un)equal distribution of resources) emerge. Some assert the role of factor endowments, in the form of predisposition to cash or food crops. An unequal resource distribution would emerge in areas more suited to cash crop production, whereas more egalitarian institutions would be characteristic of areas suitable for food crops. 6 Others point to the feasibility of European settlement, claiming that where large-scale settlement of Europeans was possible, more egalitarian institutions emerged. 7 From a historical perspective, one key contributor to economic inequality is land distribution. Land is not only a source of livelihoods but also a source of rents. Rents can be accumulated into wealth and passed on through generations. 8 Furthermore, land may operate as collateral in capital lending, further deepening inequality. 9 It is therefore not surprising that the inequalities stemming from inegalitarian land distribution tend to persist over time and contribute to wealth inequality. 10 Several studies have investigated both the drivers and the consequences of land distribution across time and space. For North America and Europe, the evidence on land inequality and its long-term effects is vast and growing. Land distribution was fairly inegalitarian, yet inequality was comparatively lower in some newly founded settler colonies in North America. 11 South America has also received attention from a historical perspective. The general picture is that land distribution in South America was highly inegalitarian in a comparative perspective. 12 For other regions of the world, we still know very little. This is particularly true for Africa, where the lack of good-quality historical sources continues to limit research. In recent years, a growing body of literature has focused on historical income inequality in Africa. 13 In contrast, only a few rare studies have examined wealth inequality in colonial Africa. Fourie and von Fintel analysed wealth inequality in the Cape Colony during the seventeenth and eighteenth centuries, finding that inequality was high from the onset and increased over time, but their study does not provide information on land. 14 Galli and Rönnbäck have examined the distribution of wealth in the Colony of Sierra Leone for one single benchmark year, which does not make it possible to study the development of the distribution of land and other resources over time. 15 Relying on a newly assembled dataset from archival sources for the Colony of Sierra Leone, we examine how the distribution of land evolved over time from the late eighteenth to the early nineteenth century. These data provide insights on where the roots of land inequality may lie. To the best of our knowledge, this is the first study to examine the evolution of land inequality over time at such an early stage in West African history. The Colony is an important case study for inequality in colonial Africa. Not only was Sierra Leone the first British colony founded in West Africa, but egalitarian ideals also played a major role in its early history, allegedly informing land distribution. Hence, the Colony is a good case study for assessing the effect of institutions on inequality.
Two major findings emerge from our data. First, land distribution was initially highly egalitarian, compared to estimates for settler colonies in North America. We suggest that this may be the result of egalitarianism being institutionalized, so that all settlers were provided with a certain amount of land upon arrival. Second, we find that land inequality was substantially higher by the end of the period studied. The increase corresponded to a policy change that allowed settlers to appropriate land more freely than was possible earlier in the Colony's history. This change created an opportunity for some households to appropriate more land than others, contributing to a more unequal distribution of land. I The territory of the Colony of Sierra Leone approximately corresponds to what is now known as the Western province of Sierra Leone or the Freetown Peninsula. The territory had been in contact with Europeans for centuries before becoming a British colony, thanks to its involvement in the commodity and slave trades. 16 The region furthermore attracted the attention of several European powers, including Britain, interested in the favourable characteristics of its natural harbour, which was the most suitable site along the coast of West Africa for a navy base. 17 Yet for a long time the high mortality rates of Europeans made it impossible to establish a European settlement in the area, limiting the interaction to a small trading post. 18 The area was deemed too unsafe even for the settlement of British convicts, particularly after the failed colonial experience in Senegambia. 19 Nevertheless, the plan for a settlement in the area would soon be revived.
After the American Revolutionary War and the Somersett ruling of 1772, 20 the English government granted freedom to some groups of slaves both in the Old and the New World. 21 A victory for the abolitionist movement, it soon became apparent that the ruling triggered social problems, collectively known as the 'Black Poor' problem. 22 One critical issue was where to settle these newly freed individuals. The Black Poor Committee (later the Sierra Leone Company) suggested resettling these former slaves in Sierra Leone in an attempt at 'humanitarian imperialism '. 23 In 1787, the first 300 settlers sailed to the 'province of freedom', as it came to be known. Upon arrival, a tract of land was supposedly purchased on behalf of the Crown from local king Tom, and the first settlement initiated. 24 Although Sierra Leone had much in common with the utopian settler colonialism that had begun with the colonization of North America, the Colony differed substantially in that its settlers were not whites. 25 The Colony of Sierra Leone became the first safe haven for liberated slaves in Africa, three decades before the Liberian experiment. 26 The Colony was to become a model of socio-economic and ideological reforms in Africa aimed at the eradication of the slave trade through 'Commerce, Civilization and Christianity'. 27 The concept of 'Bible and Plough' assumed a different meaning in Sierra Leone than in white settler colonies. There, the relationship between colonizers and blacks centred on labour (or the exploitation of it). 28 In Sierra Leone, instead, the distribution of land was intended to give rise to a class of black yeomen and to a black settler society, 29 meant to produce for the 'legitimate trade' that, according to anti-slavery activists, would supplant the slave trade and boost the Colony's development. 30 According to Granville Sharp, the man behind the foundation of the Colony of Sierra Leone, 31 two elements were required to foster the development of a successful black settler colony. The first element was self-government, with settlers sharing the burden of administration and government equally. 32 The other 22 Dresser, Slavery obscured. 23 Frenkel and Western,'Pretext or prophylaxis?',p. 212;Everill,'Experiments',Caulker,'Legitimate commerce',; Goddard, Handbook of Sierra Leone. This approach to race would again seem to re-affirm white superiority by allowing for a black settlement in an area deemed unsuitable for whites; Christopher, '"Disgrace to the very colour"'. 24 Fyfe, History of Sierra Leone, pp. 19-20. Indigenous land practices did not contemplate private property; disagreements on this aspect led to conflicts that marked the first years of the Colony. 25 Bruce, 'Utopian justifications'. 26 Everill, 'Experiments'. At the time, Afro-American return migration to Africa had only recently been proposed as a solution to oppression and segregation in the US; Pease and Pease,Black Utopia. 27 Abasiattai, 'Search for independence'; Lovejoy and Schwarz, 'Sierra Leone', p. 15; Schwarz, '"Just and honourable commerce"', p. 2. 28 Wolfe,'Land,labor and difference',p. 867. 29 The belief in the salvific nature of agriculture for blacks would spread through the empire, as apparent from the words of William King, who in 1848 in Canada wrote: '[f]rom what I know of the negro character and from what I've seen of those settled in Canada, I am convinced that the only way to improve their condition, and elevate their character, is to place them on land, give them an interest in soil and provide them with a Christian education'; reported in Pease and Pease,Black Utopia,p. 19. Isaac Land furthermore argued that 'the anti-slavery movement did not benefit from free blacks living in Nova Scotia; they needed the tropical location where they could prove that free labour was capable of raising tropical products'; reported in Christopher, '"Disgrace to the very colour"', n. 69. 30 Pybus, '"Less favourable specimen" ',p. 98;Everill,'Contested perception';Olabimtan,'Church Missionary Society projects',p. 203;Curtin,Image of Africa,vol. 1,p. 125. Although intended as substitutes, slave and legitimate trade often occurred simultaneously during the early nineteenth century; Scanlan, 'Rewards of their exertions', p. 132. 31 Granville Sharp was not only a fervent anti-slavery activist, but he also was a devout Christian and constitutional reformist; Wallace, 'Granville Sharp'. Besides his contribution to the foundation of the Colony of Sierra Leone, he fought slavery on English soil by bringing numerous cases on slavery to court. His efforts against slavery paved the way for a growing anti-slavery movement in English society that would lead to the slave-trade ban first and the abolition of slavery later; Lyall, Granville Sharp's Cases on Slavery. 32 Everill, 'Experiments'. Self-government was short-lived, but sufficient to put forward concepts of egalitarianism among the settlers. element was land distribution. To avoid conflicts over resources among settlers that could undermine the efforts at establishing a black yeoman society, land was to be distributed among settler households according to a principle of equality of outcome. 33 Land was cleared collectively by settlers and military personnel, and the cleared land was then distributed to the settlers. 34 It was argued that only egalitarianism could lead to freedom, independence, and order, the ideal characteristics of all yeoman settler societies of the time, while at the same time contributing to the fight against the slave trade. 35 If egalitarianism was to form the basis for the distribution of land among the settlers, that did not include the indigenous populations. On the contrary, the Mende, Temne, and Bullom-the indigenous populations in the region-all faced land dispossession. They had considered the colonizers' purchase of land as a mere lease, whereas the colonizers intended it as a perpetual transfer. 36 Realizing the difference, the indigenous groups actively attempted to re-appropriate the land they had been dispossessed of for over two decades, only to be eventually defeated and forced to relinquish the entire peninsula as a condition for peace. 37 The first few years of the Colony were dramatic. Despite high hopes, only a handful of settlers survived more than a couple of years. 38 Many of them died of malaria just after landing, while others abandoned the settlement to join slave traders. 39 A few years later, a mix of coerced and voluntary settlers would begin landing in the Colony, forming the basis of a new, strongly multicultural society. 40 In 1792, the Nova Scotians arrived, a group of former slaves from the American colonies formerly settled in Nova Scotia by the British authorities at the end of the American Revolutionary War who petitioned to be removed from Canada to Sierra Leone. In 1800, it was the turn of the Maroons from Jamaica, a group of runaway slaves who were forcibly removed to Sierra Leone. 41 Following the slave trade abolition in 1807 and the role assumed by Sierra Leone as the main anti-slavery base in West Africa, other groups were landed in the Colony: liberated slaves and discharged soldiers. 42 The magnitude of this latter wave of immigration outnumbered that of any previous waves by tens of thousands of individuals. 43 By the time the latter groups began arriving, Freetown had become increasingly populous and had emerged as a major trading port along the west coast of Africa, 33 Cox-George, 'Direct taxation'. By 'equality of outcome', it is here intended that land was to be distributed to individual members of society only according to gender and age group. Households that looked alike in terms of composition would therefore receive the same amount of land. 34 Settlers would pay for the land and for the assistance received from the colonial authorities by 'donating' two days of their week to work for the Colony; Fyfe, History of Sierra Leone, pp. 44-5. 35 Goodall, Invasion to embassy. 36 Dorjahn and Fyfe, 'Landlord and stranger', p. 396. 37 Schwarz, '"Just and honourable commerce"', pp. 25-6. It was during these years of war with the local populations that the Colony had to resort to trade for survival, developing trade relations centred around foodstuffs; Herrmann, '"King"'; Misevich, 'Sierra Leone hinterland'. 38  while the space in its immediate hinterland had grown increasingly contested. 44 Sources agree that the Nova Scotians and Maroons had abandoned agriculture by the early 1800s. 45 The authorities realized that the land left untended by the Nova Scotians and the Maroons could be used for the liberated slaves, easing the issue of land distribution. Yet the original grants had no clause of forfeiture in case land remained untended, and the original settlers had no apparent interest in selling their land. 46 As a result, the colonial authorities began encouraging the establishment of villages in the interior of the peninsula where land was available; this was an attempt to solve the issue of land distribution and to relieve Freetown of demographic pressure, while strengthening control over newly acquired territories. 47 Providing the new arrivals with pre-cleared land plots proved to be too daunting a task for the weak local authorities. It is not unlikely that racial prejudice among the colonial authorities against liberated slaves also played a role. 48 As the Colony expanded with the influx of liberated Africans, the colonial authorities refrained from allocating any fixed acreage to the new settlers. 49 The liberated slaves instead were distributed throughout the Colony and provided with resources (clothing and tools) for their first three months in the country. During that time, they were expected to begin building their own houses and to clear plots of land of their own choice. 50 One of the reasons for establishing the Colony of Sierra Leone, in the opinion of English travellers and philanthropists, had been the supposedly high fertility of its soil that was evident in the large quantities of crops being traded at Freetown. 51 Agriculture was intended to fuel the legitimate trade and help the Colony grow. However, it did not take long to realize that the soil of the Colony was not as good as expected and that much of the produce came from neighbouring regions. 52 In fact, the areas along the coast suffered from aridity and lack of nutrients, and the soil in the interior was rocky and easily eroded. Whereas the original settlers soon abandoned agriculture in favour of more lucrative activities in trade and administration, 53 later settlers devoted only a negligible share of their land, approximately 5 per cent, to producing for legitimate export or for the Freetown 44 It is undisputed that a large portion of the Colony's wealth, particularly urban wealth, was the result of trade. Sierra Leone was at the centre of an extensive network of trade routes extending to both near and far hinterlands, which contributed to raising Freetown to prominence for trade in West Africa. The historiography on the importance of commerce for the Colony's survival and development is extensive; see, for instance, Scanlan market, with the lion's share being devoted to subsistence agriculture, to the despair of the colonial authorities. 54 Rice, cassava, yams, and cocoa were the staples of domestic production, yet productivity was low, and a surplus was a rarity in the Colony. The vast majority of the exports from the Colony comprised produce from the hinterland, at the time lying outside the jurisdiction of the Colony. Re-exports focused particularly on palm oil, timber, and rice. 55 Exports from the Colony were limited and focused on ginger, groundnuts, pepper, and gum copal, much of which could be simply harvested without significant investment in cultivation. The production of cash crops for export, such as cotton and coffee, never became popular in the Colony. 56 Although the legitimate agricultural venture may have been a fiasco, the principles of equitable resource distribution appear to be in total opposition to the struggle for access to resources that would characterize twentieth-century Sierra Leone, making the early colony of Sierra Leone a valuable case for studies of inequality.

II
In this study, we are interested in land distribution and how land inequality evolved over time in the Colony of Sierra Leone. For this purpose, we have assembled data on settler households' landholdings for three benchmark years: 1792, 1800, and 1831. Ideally, we would have wanted to extend our study further in time. However, no further data were collected by the colonial authorities until the late nineteenth century. 57 Data for the first two benchmark years have been assembled from the Ordinances of the Settlement of Sierra Leone, collected in several volumes by Algernon Montagu. Among the ordinances are documents reporting how land was distributed among settlers arriving in 1792 and 1800 who settled in the area of Freetown. 58 Data on the size of landholdings, name of proprietor and specific location are listed for the first 498 Nova Scotian households settled in the Colony permanently in 1792, as well as for the 112 Maroon households settled in 1800, respectively. Because nearly all previous settlers had perished or abandoned the Colony and the indigenous population lived outside the border of the Colony, the data for 1792 cover virtually the entire population of the Colony at the time. This is not the case for the Maroons sample for 1800, due to the fact that a number of Nova Scotian settlers still lived in the Colony at the time. This second sample must, therefore, be interpreted with caution. These sources are nonetheless the only surviving pictures of how the Colony appeared before the large inflow of recaptives 54 TNA, CO 267/91, Report of the Commissioner of Enquiry 1825-1828, fos. 31-2; 73-6. 55 TNA, CO 267/172, Report of the Commissioner of Inquiry on the West Coast of Africa-Sierra Leone, 1842, p. 17. 56 The sources mention only a couple of small-scale attempts ever being made at cultivating cotton and coffee in the Colony. As it happened, both were unsuccessful, and the land soon reverted to subsistence agriculture; TNA, CO 267/91, Report of the Commissioner of Enquiry 1825-1828, fos. 73-6. Agriculture became profitable in Sierra Leone only after annexation of the Protectorate's fertile plains in the second half of the nineteenth century. 57 Notes reported in the Blue Books suggest that the authorities deemed data collection unmanageable; Everill, 'Contested perception', p. 185. 58 Montagu, Ordinances, vol. 3, app. 3 'Allotments'. began less than a decade later. 59 This source, unfortunately, does not inform us about household size and composition. To analyse how important these were with respect to land distribution, we rely on two additional sources: the 'List of those blacks in Birchtown who gave their names for Sierra Leone in November 1791' referring to the Nova Scotians, and the 'List of Maroons as returned by their Superintendent in April 1802'. 60 The former list only provides information on those Nova Scotians residing in Birchtown. 61 The households that we are able to link between the sources constitute a relatively substantial share of all households found in Montagu's Ordinances and do not look dissimilar from those that we are not able to link. 62 The latter list is part of the very first census attempted in the Colony, and it does not seem to pose major challenges in terms of reliability. 63 Both lists provide similar information on the name, age, and profession of each head of household, along with an account of the rest of the household members by gender and age group. The source for the third benchmark year is the 1831 Census of the Population of the Colony of Sierra Leone. 64 This census is an extraordinary snapshot of the Colony providing household-level data covering the entire settler population, organized geographically by urban/rural areas. The data show that in 1831, the majority of the population (70.8 per cent) lived in the countryside, a pattern dating back to the 1820s. 65 The geographical distribution of the population was a result of the foundation of the numerous settlements throughout the peninsula by Governor Charles McCarthy (1816-1824), 66 as shown in figure 1.
The rural part of the census includes information on the geographic location of the households; the demographics of the population; household composition; occupational titles; and key assets owned by the household. The latter gives account of three types of assets: landholding, housing type, and livestock. These, combined, made up the largest share of wealth in pre-industrial rural societies, and similarly in rural Sierra Leone. 67 The data on landholdings refer to the physical size of the households' landholdings, and not to the economic value of the plots. 68 The geographic location of the households distinguishes between 37 villages, all of which belong to one of nine rural districts.
The 1831 census was the result of a thorough and organized effort by the colonial authorities to conduct a reliable and comprehensive survey of the entire population residing in the Colony. The aim was to provide better for the colonial 59 The term 'recaptive' defines those enslaved individuals who were freed from slave ships and landed at Freetown in the aftermath of the abolition of the slave trade in 1807. 60   population, particularly for the liberated slaves, and to counteract the phenomenon of re-enslavement that plagued the country in the early decades of the nineteenth century. 69 The census takers employed standardized forms in their journey through the country to avoid discrepancies in the type of information and manner in which it was recorded. 70 The records were personally surveyed and compiled by the officers, except for landholdings, which were self-reported by the head of the household. The census does not report information on when the members of the household arrived in the Colony; thus, we cannot distinguish new settlers from old ones.
The 1831 census recorded land only in the rural districts of the Colony, but not in Freetown. Town and country lots were, however, different. The former were of a standard size of 0.084 acres and were intended for a house and a small backyard, rather than for proper agriculture. 71 Country lots instead were expected to accommodate agriculture and livestock grazing as the main form of subsistence. 72 Land is reported in 28 out of 37 villages, for a total of 5,585 households (approximately 80 per cent of the entire rural sample). 73 For those observations that are missing, the census takers claimed that it was impossible to ascertain the correct quantity of land each household occupied, but they argued that, on average, it may have amounted to three acres per household. 74 It could be argued that issues might arise from the self-reporting of landholdings if settlers feared they could have to pay a quit rent on land, as had previously occurred. The colonial administration had attempted to exact a quit rent on land in the late eighteenth century but soon stopped, following a failed rebellion. 75 By 1831, the quit rent on land had been dismissed and the matter settled. 76 Besides conscious under-reporting, self-estimates of acreage might be problematic in terms of reliability and consistency. The commissioners did certainly cast doubts on the ability of the settlers to estimate the size of their own land. 77 However, it seems improbable that heads of households vastly over-or under-estimated their landholdings by mistake.
Household-level data for the three benchmark years (1792, 1800, and 1831) are employed to study land distribution and the corresponding inequality levels in landholdings over time. Ideally, we would have liked to link information for individual households across the three censuses. Unfortunately, name changes, spelling differences, and non-unique names are common issues that make this type of venture impossible. 78  between the censuses at an aggregate level. Conventional summary measures of inequality, such as Lorenz curves, Gini indices, and percentile ratios, are used for this purpose. 79 III Previous research has suggested that several factors are associated with land distribution and could explain differences in inequality over time. One factor that may affect between-household inequality is household size. Whether because land redistribution accounted for household size, or because larger households could devote more labour to clearing land, household size could contribute to explaining inequality levels. Secondary sources claim that Nova Scotian households would receive four acres for each adult man, two acres for every adult woman, and one acre for each child 80 -yet the extent to which this policy was adhered to remains to be uncovered. The same distributive policy was not necessarily in place in 1800, and was definitely no longer in place by 1831. We first test whether the households arriving in 1792 actually received the amount of land they were officially promised. We then test the relationship between household size and landholdings for all three benchmark years by employing a regression through the origin with household landholding as the dependent variable and family size as the independent variable. By removing the constant, it is possible to test whether the distribution of land is associated with household size. 81 As mentioned, a shift in the policy governing land distribution occurred during the period under study. If the first settlers received a set amount of land, as a means to ensure equal participation in the colonial agricultural venture, later everyone could appropriate and clear as much land as they could manage, opening opportunities for appropriation that did not exist before. This approach could have had important implications for the nature of settlements in the Colony because it enabled specialization to a higher degree than was previously possible.
A second factor that might explain inequality levels is the time since settlement, because this was a recently established settler Colony inhabited by settlers arriving in different waves. One could expect that households that arrived earlier would have had more time effectively to appropriate larger portions of land than households arriving later. Unfortunately, our sources do not inform us of when individual households arrived in the Colony. We thus rely on when individual villages were established as a proxy for the settlement of individual households in the village. 82 Third, factor endowments across the Colony might have influenced land distribution. 83 If a district was suitable for agriculture, it might have been more attractive to specialize in agriculture and appropriate more land there than in areas where land was comparatively less fit for agriculture. In this regard, an estimation 79 Jenkins and Jäntti, 'Methods'. 80 Cox-George, Finance and development; Walker, Black loyalists. 81 Eisenhauer, 'Regression'. 82 The year of settlement was extracted from TNA, CO 267/111, Census of the Population of the Colony of Sierra Leone, 1831; Fyfe, History of Sierra Leone. 83 Engerman and Sokoloff, 'Factor endowments'; Sokoloff and Engerman, 'History lessons'; Easterly and Levine, 'Tropics, germs and crops'; Clark and Gray, 'Geography is not destiny'. of inequality in terms of land values, rather than with respect to surface area, would have allowed us to control for this aspect, assuming that the land market functioned well. However, our sources do not provide information on the value of the land held by the households, and our efforts to find alternative sources of land prices in the rural part of the Colony were not successful. 84 We have therefore attempted to include factor endowments in the analysis by relying on soil quality data by village. Settlers had been geographically concentrated in the area of Freetown in 1792 and 1800. The ecological endowment was thereby rather similar for most plots distributed to the early settlers. This was no longer the case in 1831 when settlements were more dispersed geographically, and factor endowments therefore more diverse. We exploit this element to create a crude proxy for soil quality in each settlement. The 1827 Report of the Commissioner of Enquiry provides brief information on soil quality at the village level. 85 The Report mentions large differences in soil quality throughout the peninsula. Due to the importance of such information for our study, we check the reliability of this information against a modern-day source of information on soil quality for Sierra Leone, the Food and Agricultural Organization's report 'Land in Sierra Leone: a reconnaissance survey and evaluation for agriculture '. 86 From this source, we extract data on soil quality for the Sierra Leone peninsula at the village level. 87 The land in the Colony can be categorized into three distinct soil systems. The first system is typical of the mountainous areas in the centre of the peninsula, which extends to the seashore for most of the western part of the peninsula. Most of the land is located on steep slopes covered by thick forest. Once the forest is cleared, soil fertility is short-lived and soon exhausted due to high erosion levels. The second soil system covers a limited part of the peninsula but has somewhat better soil quality. Typical of this system are terraces where continuous cultivation of rainfed crops can take place, and swamps suitable for paddy rice cultivation. The third, and best, soil system primarily extends eastward from the mountains towards the hinterland. The land comprises low-altitude floodplains where erosion is limited, and the soil is deeper and less gravelly than in the rest of the peninsula. 88 The information found in these sources was used to categorize each village on a scale from 1 to 3, with 1 being the worst soil quality and 3 being the best. 89 Fourth, geography may have played a role. By 1831, the port city of Freetown had become a substantial market for goods and services, as well as a large factor market for labour. 90 Villages located close to Freetown would have had an incentive to supply these markets. A variable for distance to Freetown was constructed by georeferencing the location of the original settlement and calculating the geographical 84 A legal market for land had been established in 1796, yet no trace of a land market outside urban Freetown exists in the sources we have consulted. 85 TNA, CO 267/91, Report of the Commissioner of Enquiry 1825-1828, fos. 5-43; similar brief evaluations also appear in a later report: TNA, CO 267/172, Report of the Commissioner of Inquiry on the West Coast of Africa-Sierra Leone, 1842. 86 Birchall, Bleeker, and Cusani Visconti, Land in Sierra Leone. 87 We assume for this study a high degree of historical persistency in the type and quality of the soil on the peninsula. 88 Birchall et al.,Land in Sierra Leone,; TNA, CO 267/91, Report of the Commissioner of Enquiry 1825-1828, pp. 5-43. 89 See online app. S1 for summary statistics. 90 Galli, 'Socio-economic status'. distance in that manner. 91 However, modes of communication would also have mattered significantly in terms of market access. Contemporary sources affirm that the most common means of transporting goods to the market-mostly Freetownwas that of head porterage, whereas the use of donkeys was not widespread. 92 Another important means of transportation was by canoe. The Report of the Commissioner of Enquiry, along with a map produced in 1827, provides information on routes of communication between the villages and the capital and on their type and quality. 93 This information was used to categorize each village on a scale from 1 to 3, with 1 having the worst communication quality and 3 having the best. 94

IV
The first aim of our study is to establish how land distribution evolved over time.
Our first data point, 1792, corresponds to the first extensive land distribution that took place when Nova Scotian settlers landed on the shore of Sierra Leone. The second point refers to the distribution of land that took place when the Maroons from Jamaica arrived in Sierra Leone in 1800. Our final point in time is 1831, and it corresponds to the land distribution recorded in the first comprehensive census taken in the Colony. The Colony had by then extended to include the entire peninsula of Sierra Leone, where approximately 20,000 liberated slaves had been settled. Table 1 shows data on the distribution of land by household for the three benchmark years. The mean allotment of land in 1792 amounted to 5.2 acres. At this stage, only a handful of the first settlers would have still lived in the Colony, effectively meaning that the 1792 sample was close to the total population in the Colony. The distribution of land among these settlers was relatively even: whereas no household held less than two acres of land, the largest amount, which was held by a few households, was 11 acres.
The distribution in the 1800 sample of Maroons shows a different pattern. The landholdings ranged from 0.33 to four acres, a much smaller range than that of 1792. The Maroons held substantially smaller plots than the Nova Scotians, on average only 1.33 acres. Although no Maroon household in the sample was left completely landless, almost half of them owned less than one acre, and only three households held more than three acres.
In the 1830s, the distribution of land was significantly more unequal among the rural households in the census than it was in the two previous benchmark periods. The mean allotment size hovered at three acres of land per householdlower than the Nova Scotians but higher than Maroons-yet 9.3 per cent of the total owned no land at all. On the other hand, a small share of households owned considerably larger plots of land, with the largest single possession reported at 90 acres, owned by Edward Prince, a discharged soldier who had managed to become a large landowner. 91 The path of the main roadway crossing the peninsula of Sierra Leone and connecting most of the area in our study has changed very little over the last two centuries. We thus assume that the original pathway, depicted in Montagu, Ordinances, app. 3, has persisted to the modern day. 92    The measures of inequality reported in table 2 and figure 2 all suggest that land inequality increased over time in Sierra Leone. The findings of increased inequality, with the estimated Gini coefficients increasing from 0.258 to 0.453, are supported by an examination of percentiles and percentile ratios. The share of land owned by the top 5 and 10 per cent increased by more than 10 per cent over the period. Furthermore, the top decile saw a staggering threefold increase in land owned in relation to the bottom decile between 1792 and 1831, probably due to the large number of landless households in the latter sample. The top decile doubled the ratio of land held even in comparison to the 50th percentile, whereas the bottom decile lost some ground in comparison to the latter group over time.
Despite an increase in land inequality over time, inequality in the Colony remained comparatively low from a global perspective. Although no existing   as official policy stipulated at the time. 96 We find that half of the Nova Scotian households that we were able to link received an amount of land comparable to what they were entitled to according to their family composition. The remainder received less land than promised, on average 1.6 acres less. 97 The discrepancy in the amount of land received could be explained by mortality because data on household size for the Nova Scotians pre-dates their trans-Atlantic crossing. A high price in lives was paid during the crossing of the Atlantic and even more so during the first few months in the Colony, with the latter period's death rate estimated to be at least 17 per cent. 98 Fewer individuals were thus alive when the distribution of land took place in the Colony than before their departure from Canada, potentially explaining why many plots seem to have been smaller than official policy stipulated. Table 4 shows that land distribution in 1800 still depended on household size, although land had been distributed to the households on a smaller scale than in 1792 despite the larger average household size. The results suggest that household size also mattered for land distribution in the 1830s. However, the explanatory power appears substantially diminished compared to the two previous benchmark years, presumably indicating a different mechanism governing land distribution. In total, however, the evidence seems to suggest that if inequality rose over time as our findings so far suggest, it was not due to differences in household size. In 1831, household size was both smaller on average and less widespread around the mean than it had been in the two previous benchmark years. Table 5 expands on our previous analysis to include, alongside household size, other factors that may also contribute to explaining land distribution and land inequality over time using a pooled regression of the three benchmark years. These factors are: time since establishment of a settlement, as a proxy for time available for land appropriation; factor endowment, captured by soil quality; and geography, in the form of distance and communication quality with the capital. The first model examines only the main effects, whereas the second model introduces an interaction term between household size and years since establishment aimed at examining whether the effect of household size on acreage differed for different values of time available for accumulation. The results in table 5 suggest that household size had a positive and significant effect on acreage, implying that larger families held on 96 Cox-George, 'Direct taxation'. 97 1.6 acres equals the amount for one-and-a-half children, whereas it is only slightly lower than that for an adult woman. 98   average larger plots. However, the effect drastically increases in model 2 because this represents the effect when the year since establishment was 0 (that is, 1792).
In model 1, we find that on average across all censuses, the number of years since the establishment of settlements has a mildly negative but significant effect on landholdings, corroborated by evidence suggesting that recently arrived settlers on average appropriated more land than older settlers (see online appendix figure S2). In model 2, when interacting the effect of household size and time available for land appropriation, we find that the interaction factor is negative and significant, and the main effect of years since establishment has become positive and significant. This result suggests that the effect of household size differs for different levels of years since establishment. The effect of the interaction is mildly negative, meaning that in older settlements, household size had a lower impact on household acreage relative to younger settlements. Taking into account that the younger settlements correspond for the most part to the earliest censuses (1792, 1800), whereas the older establishments refer to the latest census (1831), we can conclude that the impact of household size on landholdings was lower in the latter period than in the previous two samples (1792,1800). 99 The effect of geography, in the form of distance from Freetown, does not have any significant effect on landholdings. 99 The effect can be better visualized in online app. fig. S1. Therefore, distance does not contribute to explaining land distribution nor inequality trends. Although the first two settlements (1792, 1800) were established in the area of soon-to-be Freetown, more recent villages were founded farther away but without a clear-cut pattern (online appendix figure S3). This expansionary settlement pattern, however, meant that land was initially concentrated in an area of approximately 10 km 2 , whereas the area across which the Colony extended by 1831 had grown to roughly 550 km 2 . Geographic differences were therefore more marked in the latter benchmark year than they had been previously. 100 The results from table 5 show that the effect of soil quality on plot size is positive and significant, suggesting that the effect is stronger where the levels of soil quality are better. As it appears, villages founded at a later date had been established in comparatively more fertile areas than earlier ones, even relative to Freetown itself (see online appendix figure S5). The effect of communication quality on the size of landholdings, on the other hand, is negative, although its significance is limited to the 'good' level, and insignificant otherwise. Despite the effect of individual factors, the R 2 -coefficients of the regressions in models 1 and 2 are able to explain only a fraction of the variation in the dataapproximately 15 and 17 per cent, respectively. Although the model's explanatory power improves by adding the interaction effect, suggesting that the mechanisms behind the changes in land appropriation over time indeed were associated with household size to a certain extent, the R 2 -coefficients are nonetheless lower than if the model had been run separately for each benchmark year. 101 Therefore, the variables in models 1 and 2 explain marginally less in each benchmark year, supporting the hypothesis that other factors played a role in explaining land distribution in the latter period and the trend of increasing inequality over time.
One such element could be that factor endowments and geography may have influenced inequality via occupational specialization. If low fertility rendered it difficult to rely on agriculture for one's livelihood, settlers may have been incentivized to engage in other occupations. Table 6 shows that the First Mountain and Kissy areas exhibited the worst soil quality among all the districts in our study.
In First Mountain, settlers seemed to have specialized out of agriculture: average landholdings were small (less than two acres per household), and only about a quarter of the households' heads were recorded as farmers. The remainder were occupied in manual skilled and unskilled occupations, such as sawyers and different types of construction workers. In Kissy, instead, agriculture was the primary occupation of over 80 per cent of the household heads. Average landholdings in this district were, however, also comparatively small (on average two acres per household), and the inequality in landholdings between households was likewise small. Despite its limited suitability for agriculture, qualitative sources report that Kissy was the only district involved in agricultural production for the Freetown market to any substantial extent. 102 Our data seem to indicate that people in the district of Wellington might have been involved in agricultural production for the market as well: the vast majority of the household heads were classified as farmers, average landholdings were comparatively large (on average four acres), and the people living in this district would have had reasonable access to the market in Freetown. This finding is corroborated by qualitative evidence that indicates that Wellington did partake in commercial agricultural production, with a focus on cassava and maize, although to a more limited extent than in Kissy. 103 The levels of inequality in landholding were extraordinarily high in this district, with a Gini index as high as 0.63. It would therefore seem as if some households had been able to appropriate very large tracts of land, possibly to participate in commercial agricultural production for the Freetown market.
The remaining districts-Waterloo, Hastings, York, and the Banana Islandall exhibited average or above-average soil quality, which could have created an incentive to specialize in agriculture in these districts. Average household landholdings were also comparatively large, in the range of three to four acres per household, yet access to the Freetown market was not as good as in other parts of the peninsula. Commercialization of the local economy in several of these districts appears to have been limited, with only a small number of traders and shop owners recorded, and the vast majority of them residing in the district of York in particular (see online appendix table S4). To the extent that the heads of households were recorded as having agriculture as their primary occupation, such as in the district of Waterloo and to a slightly lesser degree on the Banana Island, this was probably as subsistence rather than commercial agriculture. This difference might contribute to explaining why the early Colony of Sierra Leone failed to develop into a major exporter of agricultural crops, despite initial expectations that it would.
In Hastings and York, the core of the household heads were recorded as occupied in non-agricultural occupations, predominantly as unskilled labourers, sawyers, and carpenters. Qualitative sources affirm that a flourishing timber industry for export had been established in Hastings soon after the area had become part of the Colony, providing a lucrative alternative to agricultural production that is reflected in the tiny fraction of heads of households recorded as farmers. 104 York saw the co-habitation of trading activities, thanks to its favourable position in relation to the 102 TNA, CO 267/91, Report of the Commissioner of Enquiry 1825-1828, p. 5. 103 Ibid., p. 40. 104 Fyfe, History of Sierra Leone, p. 125.
Grain Coast just south of the peninsula, with sawing and building. To the extent that the comparatively large acreages of land owned by the households in both these districts were farmed, it ought to have been other members of the household who undertook this activity.

VI
This article has examined how land distribution evolved over time in the Colony of Sierra Leone at a particularly early stage of African economic history. The Colony provides a unique case study for the analysis of inequality because it was allegedly founded upon egalitarian ideals in the late seventeenth century. Our findings suggest that these ideals were actually implemented in the form of land distribution to the first settlers arriving in the Colony in 1792. This result makes the Colony of Sierra Leone look highly egalitarian in comparison to both North American settler colonies and other African colonies. Colonial institutions could in certain contexts thus contribute to the development of comparatively egalitarian societies, even in a colony where settlers were not white. This finding casts some doubts over theories suggesting that inclusive institutions, as a rule, were established by European settlers for European settlers only. Over time, however, Sierra Leone experienced rising land inequality. By the last benchmark year under study, in 1831, the levels of inequality had increased substantially. We find that a small elite had come to appropriate substantially larger tracts of land than ever before, while in parallel, a substantial number of landless rural households had also emerged. This observation is consistent with results from prior inequality literature from other parts of the world that found a rise in inequality over time. Our analysis suggests that, in the case of Sierra Leone, the increase in inequality was enabled by a crucial shift in land distribution policy. The shift allowed later settlers to appropriate land more freely than earlier settlers, who had received a set amount of land dependent upon their household composition. The policy shift enabled inequality to increase because some households-primarily in several of the younger villages-came to appropriate comparatively larger acreages of land and to specialize in agriculture, whereas others remained landless.