Diversity and division: Digital mapping of censuses in the Swedish Caribbean, 1835–1872

ABSTRACT By examining the census records of Swedish-Caribbean colony St. Barthélemy, a GIS – Geographic Information System – can be constructed which maps the inhabitants of the harbor city of Gustavia, as well as the inhabitants’ placement within the census categories of the colonial administration. This article examines these categories and how they shifted over time, as well as maps out patterns of settlement across the city. One major finding is the division between the Catholic French-Caribbean side on the east and the Anglican Anglo- and Dutch-Caribbean side of the west, the latter of which was home to a number of merchants owning high amounts of enslaved people. This spatial division was also parallel to the political differences between the merchant elite and the Swedish colonial administration. Furthermore, this article encourages additional GIS studies of predemocratic Caribbean cities, particularly regarding spatial division between whites and free people of color.


Introduction
One of the defining aspects of St. Barthélemy, the Swedish colony in the Caribbean, was its rapidly shifting demography. Founded as a free port with freedom of religion, with the intention of attracting foreign merchants and mariners from the Americas, but also from Europe, it grew from 700 to 6,000 people from 1785 to 1800. Swedish neutrality in a time of almost constant imperial conflict created a community in the harbor town of Gustavia out of nothing. Since neutrality was practically the sole purpose for becoming a part of Gustavia, there was a constant anxiety that this community would disperse as soon as this advantage ended. 1815 marked the beginning of an enduring peace in the Atlantic and by 1878, when Sweden returned the colony to France, Caribbean neutrality was of little financial consequence.
Already before 1815 there were petitions expressing concern about migration from the colony. In a letter from the merchants committee in 1817, several suggestions were made to improve the conditions of trade. Here, they stressed the mobility of merchants. "He who has capital to improve, and talent to improve it, if shackled at home will seek employment for it abroad, under a foreign flag and a more tolerant state of things: and he will finish by following it up with his presence: because it will be his interest to do so!" 1 The island's population would dwindle thereafter for the rest of its time as a Swedish possession.
Carolyn Fick and Francine Mayer have authored important examinations of population growth in St. Barthélemy, as well as the distribution of enslaved Africans and free women of color across the island. They observe a clear difference in the composition of urban and rural households, with Gustavia having a higher degree of free people of color, as well as a higher degree of free black slaveholders, although whites were still in the majority. They also note a gender imbalance on the island, meaning larger amounts of women than men, especially women of color. Here they differentiate St. Barthélemy from other sugar colonies, which they argue were characterized by a higher degree of coercion and strict differentiation in master-slave relationships, although St. Barthélemy was still dependent on cruelty and forced labor. 2 However, there are three potential areas of improvement of their work. First of all, their studies do not examine the colony much after 1850 and the latter years of the colony, an area which remains understudied. Second, while slavery is crucial to understanding St. Barthélemy's society, larger demographic patterns can reveal economic and social forces which better illustrate the population of Gustavia and its slaveholders. Current research frames St. Barthélemy as part of a system of high frequency exchange of goods, people, and information across imperial borders, rather than as an aberration from larger sugar-planting island. 3 Demographic analysis of Gustavia in particular should take this new framework in consideration. Third, I believe that spatial analysis of these censuses, through the creation of a GIS -Geographic Information System -, can help us to further understand the social stratifications of Gustavia. Previous demographic analysis has treated the city as a homogenous unit, despite the administrators and censuses themselves drawing a clear differentiation between three areas. Addressing this would address the dearth of spatial history not only within Swedish colonial history, but within Caribbean urban history.

Caribbean spatial history, historical GIS and urban public space
The last two decades have seen a growth of historical GIS usage following what has been called the spatial turn, a reemphasis on space and place as not just physical, but social, cultural, and phenomenological concepts. 4 The increasing capacity and accessibility of digital tools for creating graphical analyses have encouraged historians to partake in this field to answer historical questions. Initially, these questions have been quantitative and positivist in their nature, but they have successively opened up to more qualitative analysis. As Ian Gregory claims, GIS is not a substitute for humanistic deep reading, but rather a complement to it. 5 Spatial analysis has also increased within Caribbean History, as described by Reena N. Goldthree. Analysis of space and mobility related to plantation life and slavery, as well as relations between and within urban spaces, have grown in the last ten years. 6 In terms of digital and spatial history, the most impressive project is the multimedia project by Vincent Brown, creating a visual narrative of the 1760 slave revolt, where the viewer can see the movement of armies and note details across a map of Jamaica. 7 There are other examples of historical GIS studies of the Caribbean, such as the study of the cultural landscape of eighteenth century St. John in the Danish West Indies or of maritime post routes between Spain and the Caribbean, but it is an area with much room for growth. 8 More crucially however, there is a lack of urban history of colonial Caribbean, at least concerning the city as an object of study in its own right. Traditionally, early modern scholars of the Caribbean have studied the plantation systems and society of the countryside, and have often regarded the city as merely a waypoint in the process of shipping sugar from the colony. Alternatively, urban slavery has been compared to rural slavery, and scholars have examined the movement of free people of color toward urban areas. The cities themselves have however rarely been examined or when examined, have been treated as homogenous objects without any particular differentiation or stratification, beyond the general presence of slavery.
In his study of Caribbean development, Potter points to the 1950s as the start of urbanization. While urban areas existed during colonialism, before and after slavery, they served primarily as shipping points and administrative centers. It was not until independence and the growth of mass tourism industries that the Caribbean city began to grow. 9 Thus, the very concept of the Caribbean city stands in contrast to the plantation economy and has consequently been absent from many studies of the early modern Caribbean. New studies on free ports and inter-Caribbean commerce, while aiming to redirect focus from the more traditional plantation historiography, still mostly treat cities as nodes in a network, instead of as spaces of social and cultural meetings, conflicts, and creations.
There are of course exceptions, such Linda Rupert's impressive work on Curaçao, the Dutch Caribbean island whose free port Willemstad saw a mass migration of various ethnicities. As the city grew in the eighteenth century, Rupert notes that new expansions called Otrobanda, Pietermaai and Scharloo, became different districts of differing ethnic compositions, and that their names changing from Dutch origins -Oversijde to Otrobandawas part of the creolization of the city. Also, she examines the different architecture and functions of buildings to show the growth of the city. Rupert thus examines the city as a living object in and of itself, instead of a representation of trade statistics or as a point of contrast between the rural and the urban. 10 Additionally thorough work has been done on the early history in the sixteenth century of Havana by Alejandro de la Fuente on, as well as Paul Niell's study on the art history of urban spaces in late colonial Cuba. 11 The anthology Atlantic Port Cities: Economy, Culture, and Society in the Atlantic World, 1650-1850 also features several articles on port cities in the Greater Caribbean, such as Kingston, Cartagena and Vera Cruz. 12 In Dispossessed Lives: Enslaved Women, Violence and the Archive, Marisa J. Fuentes posits that the city of Bridgetown in Barbados exhibited a form of social control in its architecture with public buildings and displays of encagement, humiliation and death of enslaved men and women demonstrating the risks of resistance and enforced obedience. Her study demonstrates how urban spaces within the Caribbean can be further examined. 13 Social communities in predemocratic city spaces have been a reoccurring topic of discussion regarding public spaces and political community. In The Structural Transformation of the Public Sphere, Habermas has most famously used public spaces such as coffee houses in enlightenment Europe as the point of emergence of a public sphere, in which the exchange of ideas regarding literature and politics led to a growing democratic culture. This has been challenged or augmented by many writers, who instead suggest the existence of several separate public spheres. In The Human Condition, Hannah Ardent has similarly looked to public spaces as the place where politics is not only discussed, but the political individual is created. However, both these perspectives separate public and private areas, which in many ways falls short in its description of predemocratic public life. In contrast, in her depiction of political life in eighteenth century Stockholm, Politikens hjärta, historian Karin Sennefelt argues that public spaces are created through actions and that human behavior creates a space as public or private. For example, by hosting a political rally or a ball, a home can be transformed from a private space into a public space. 14 For the Caribbean, in which balls were a regular method for creole merchants to display their status and create bonds, it is hard to argue for a clear division between a public and a private sphere. Similarly, cultural meetings such as dances were prohibited for the enslaved largely to prevent the creation of political spaces. I therefore argue that GIS census analysis of predemocratic Caribbean cities depict not merely the aggregation of a group of separate private spheres, but that the formations of these households creates networks of interactions which in turn create public spheres. Each property is a node within which inhabitants move in and out, interacting with each other, creating a common culture.
GIS studies of redlining and racial segregation in North American cities have been a hallmark example of the use of census data on urban spaces to highlight spatial differences among races and ethnicities, as well as their lasting legacies. 15 There are many examples of ways to examine urban spaces through GIS, with a variety of data, yet there are to my knowledge no studies that use GIS to examine predemocratic Caribbean urban spaces. Arguably, Anne Pérotin-Dumon's study on Basse-Terre and Pointe-à-Pitre in Guadeloupe comes closest, as she utilizes, among other things, census data to examine urban spaces and maps to discuss the Caribbean urban city space and its social, economic and demographic aspects. 16 Using census data from the local administration, I will demonstrate local spatial differentiation between 1835 and 1872 within Gustavia. In discussing the censuses, I will, however, not only discuss their value for the creation of historical GIS analysis, but additionally their value to colonial administration.

Censuses in St. Barthélemy
Colonial societies, the laboratory of scientific racism and modernity, often preceded the metropoles in keeping censuses and mapping out population growth. Here, administrators had to face questions regarding individuality, migration, and the degree to which enslaved Africans could be considered both members and property of a household. Meanwhile, in Europe, populations were rarely kept under as close scrutiny as in the frontiers of empire.
Molly Farrell points to the Sin of David in the Bible, in which King David was punished for counting the men of Israel and Judah, as an entry point of understanding this perspective. This Biblical story would feed anxiety regarding political statistics far into the early modern era, with John Milton and French lawyer Jean Bodin debating whether or not the Sin of David consisted of performing a census or if the Sin applied only for surveys of military conscription, rather than surveys of the general population. Even until 1771, population statistics were hindered in Great Britain by invocations of the Sin of David. Gunnar Thorvaldsen argues, however, that the Sin of David was less an authentic religious anxiety and more a theological excuse used by local priests to avoid doing extra work. Regardless, we can see how censuses were part of a transition from a feudal society of local control, where the population was in the interest of the parish or lord, to a modern society, where population statistics were of importance to national governance. 17 In Sweden, which has a long history of population statistics, this transition was primarily instigated by the Lutheran orthodoxy of the seventeenth century. In order to ensure that all members of households knew their Lutheran catechesis, house visits were done by priests, who kept notes on the religious knowledge of each individual of the household. Similarly, priests aided in the creation of statistics for drafting military personnel and taxation lists. In 1749, Tabellverket was instituted to compile this information into larger population statistics. Thus, religious expectations intersected with the needs of the fiscal-military state. Ironically, this method of continuous collection of statistics meant that Sweden was comparatively late in creating censuses, meaning a complete count at one given time. The first nominative census of the entire nation state was done in 1860, after the creation of Statistiska centralbyrån. 18 In St. Barthélemy, however, census were held as early as 1787. Here, the household was marked by the name of the head of household, and the number of people in it, sorted by status in the slave societywhite, free person of color, slave -, gender and age. 19 This, I would argue, is a numerical census. The difference between numerical and nominative censuses is primarily the difference between population as a set of numbers reflective of a larger collective and population as a collection of individuals within separate households of this collective. In this sense, nominative censuses reflect not just a social sphere wie es eigentlich gewesen ist (as it actually was), but also the conceptions of individuality and social stratification at the time of the census. While the head of household was named, the rest of the family and households became only numbers. We know that similar censuses were performed until 1794, and several population resumes written were written and sent to Stockholm. From the middle of the 1820s these became regular features of the governor's reports; we have resumes of eight to ten years between 1826 and 1836.
The organizational model for the Gustavia census was based on the numbering of property lots, a mapping of the city established already in 1787, according to instructions from the Swedish West India Company office in Stockholm. 20 Gustavia was divided into three areas, east, southoften called Centerand west. These areas contained different quarters, called for example Berget (The Mountain), Räfven (The Fox), or China. Within these quarters there were several lots, which were assigned a number. Thus, the census listed all property in numerical order, starting from no. 1 in the east and ending around no. 460 in the west. As more quarters were built and some lots were subdivided, creating new lots, this order became more chaotic, with roughly 650 plots in total by the 1830s.
Through the preservation of a cadastre (register), with graphical representation of the different quarters, I have been able to create a rough estimate of where the different lots were, creating designated markers for 500-550 lots. These lots have then been mapped onto a map created by government secretary Samuel Fahlberg in 1807, intended for further construction around the harbor. 21 This map was, furthermore, was made facing southwards, meaning that the area west is to the right of the map and the area east is to the left. Inherent in the methodology of using this map is of course the temporal shifts between the map and the censuses, with a span of 65 years between this map and the 1872 census. However, since the population decreased in Gustavia during this time, there was no growth in urban space that was not captured by the 1807 map. Nevertheless, the final GIS must be viewed as an approximation of graphical representation as these censuses carry an inherent degree of uncertainty due to clerical errors and choices in what information to convey and damages to the documents.
We can see from several proclamations how the censuses were performed. For example, on 18 March 1813, it was proclaimed that between the first and third of April, proprietors of houses or their representatives from the three sections of Gustavia were to report to the government secretary, first east, then south, then west. There, they would make a "Declaration of the Persons belonging to their Families, Parents, Servants or Slaves, either their own or under their care, stating their names, age, place of nativity, trade, occupation and colour." Additionally, they should procure the same information for properties they own. Failure to attend and provide information would be punished by a fine of five Spanish dollars. 22 If we compare this proclamation with the one made in 1854, there is one significant event to consider, namely the Swedish abolition of slavery in 1847. 23 Thus, the only difference is that there was no reference to slaves, and that they should state only "names, age, place of Nativity, trade or occupation," with no mention of color. However, color is still referenced in the census, as well as emancipated slaves, as will be elucidated below. The five Spanish dollar fine was the same. 24 We should, however, not assume that the proclamation completely reflects the actual census proceedings, but is rather an approximation. For example, it was not until 1860 that the proclamation stated that sex of inhabitants should be noted, despite the fact that it had been present in all previous censuses. It was simply assumed as naturally pertinent information. 25 Similarly, since the accuracy of each report depended on self-reporting, it is almost guaranteed that some households declined to report members. Masters could wish to underreport the enslaved owned for taxation purposes, runaways could be kept hidden and long-term guests could be interpreted as not official household members and not be reported to the census. Therefore, these censuses should be considered an approximation rather than an absolute number.
In studying the spatial structure in Gustavia, I have created databases of the censuses of 1835, 1854 and 1872. By reviewing the changes of the census formats, we can understand what information was considered relevant as the social and legal order of St. Barthélemy changed. In the 1835 census, the owner of the lot was noted first, next to the first member of the household. If the owner of the property lived in the household, the name was written covering both these areas. I have accordingly marked the first person listed in each household as the person presumably responsible for its affairs, which I will refer to as its primary member. Often, household affairs were much more diverse and complex, but it will enable me to simplify certain statistical patterns.
Afterwards, all other members were listed. In my database I have noted the location and owner of the property, the amount of separate households, the religious denominations present, the profession, birth place and gender of the primary member, the number of whites, free people of color and enslavedor formerly enslavedpersons present divided by three age categories, as well as if the owner and the primary member was the same and if the property was uninhabited. To a certain degree in my own GIS analysis, I have thus adapted the categorization of the colonial administration, such as at what age children, adults and the elderly were divided, as well as what the geographical distinctions were in the city and what professions people had. One could, for example, argue that I reproduce the distinctions made between free people of color and whites according to the Code Noir, which considering the class diversity within these groups is not as ironclad as one could assume. 26 To best of my ability I have tried to keep this in mind during my analysis.
The censuses are also, as many nineteenth century archival documents, messy in format and physical condition. Miscalculations, illegible writing, torn or missing pages should be factored into this documentation, beyond the inability to estimate all locations or the inherent bias of the original censuses. That being said, the sheer amount of data within the census allows us to move past specific inaccuracies in order to draw larger conclusions. At times, references were drawn to censuses a few years prior or after in order to corroborate family names or similar details. In my personal estimation, at least 95 percent of all entries into the database and included in the GIS are accurate in comparison to what was reported at the time.
Household members were divided partially by age, i.e. whether they were below 15 years of age, 15-64, 65+, and partially by their status according the Code Noir: white, free person of color, or enslaved. This created nine different points, which were marked with dots and tallied up towards the end. Also, the specific age of the person, and their place of birth was noted, as well as the profession of the primary member. In the 1835 census, professions were very sporadically noted. In the final column, additional notes, such as if someone was absent during the taking of the census, was noted.
By the 1854 census, there were two primary shifts. First of all, due to the abolition of slavery in 1847, the enslaved were no longer a category. However, there was a mark for those who were emancipated specifically as part of this legal shift. Those who were enslaved but received manumission before 1847 did not receive this mark, as can be seen with certain people born in Africa who were not counted as an emancipée. Second, the category depicting jobs were much more frequently marked, with many jobs performed by women not being displayed, such as washer, cigarmaker, trayseller, or seamstress. In this particular census, the houses damaged by the 1852 fire, in which 135 houses were burned down and 400-500 people became homeless, were specifically marked, as well as the house in which the fire originated. 27 The 1872 census had a major shift, in which the categories separating whites and people of color disappeared, while the religious denomination of the members of the household appeared for the first time. Because of this, it is difficult to find a specific link between race and religion or track potential shifts in such links from 1854 to 1872. The markings denoting emancipated citizens as part of the 1847 abolition remain, although by this time there were only a few dozen.
Based on these changes, we can see that censuses in St. Barthélemy operated according to categories based on local knowledge, hierarchies, and necessities. For example, the legal division between whites and free people of color had already ended by royal decree from the metropole in 1833. 28 However, the distinction remained, with whites, free people of color, and the formerly enslaved being referred to as the first, second and third class, even when proclamations no longer stated color as a necessary category to report. The mapping of the fire and denoting ownership of lots were also aspects of the utility of the census for local administrative information. Often, the demarcation of property would state not just its current, but former owner. Other times, they were listed as the successors or children of a former owner, perhaps as shorthand for a much more complicated matter of inheritance.
One must also consider the archival dimensions of these censuses. After the return of the colony to French dominion in 1878, there was an archival distinction made between the documents sent to Stockholm, found in the Swedish National Archives, and those remaining in the Caribbean, later moved to the Archives Nationales d'Outre Mer in Aixen-Provence. The early censuses from the late 1700s can be found in the Swedish National Archives, but the complete censuses examined in this article can only be found in Archives Nationales d'Outre Mer. After 1800, the Swedish National Archives generally contain short numerical summaries of the results, appended to the governor's reports. Thus, it is clear that the later censuses were never seen in full in Stockholm. It is questionable if there was any initiative from Stockholm in the creation, categorization, or execution of census taking at all. 29 Therefore, it is clear that the changing categorizations, with an increasing focus on professions, as well as a shift from racial classes to religious denominations, indicates changes in what was deemed as information critical for collection within the colonial administration itself. Artisanal and domestic professions became of higher importance as the colony's commercial standing decreased and slave labor ostensibly became free labor, making information on who was a household servant, carpenter, or seller of local produce more relevant. Birthplace was always a category present within the censuses, indicative of the importance of various geographical origins for local administration of a metropolitan population, but it was omitted from the summaries.
It is within this local framework that notions of racial classification must be understood. When we compare the racial categories of the censuses to those sent in summaries to Stockholm, following the abolition of legal distinctions between free men, we can see a variety of approaches. In the summary of the 1835 census, the racial categories are "blancs," "de couleur libres," and "esclaves," while for the 1836 summary, "1er classe," "2eme classe," and "3eme classe." 30 For the 1840 census summary, there are two main categories: "libres" and "esclaves." However, the "libres" category is divided into two subcategories: "1er classe" and "ci-devant 2eme classe." 31 Free people of color has thus become the category formerly categorized as second-class citizens. This is not a change seen in the original 1840 census, where they are still referred to as the second class, and by 1843 free people of color are again referred to as the "deuxieme classe." 32 By 1854, the categorical distinction between whites and people of color is recorded in the census, but omitted from the summary. 33 These changes was not the result of conceptual distinctions between different administrators, as all the mentioned summaries and censuses from 1840 to 1854 were written and signed by the same person, S. A. Mathews. 34 It is clear that even after royal decree, free people of color were still regarded as second-class citizens and that almost no efforts were made to hide this distinction. T Thus, racial classification operations did not emerge from the metropole, but from the local power dimensions and stratification within the colony and within Caribbean society. This should not be interpreted as different understandings of race, but rather of the relevance of race to the given society. Stockholm did not care particularly one way or another how many in St. Barthélemy were free people of color, but St. Barthélemy cared enough to keep counting even after they had stopped telling Stockholm.
In my examination of these censuses, I will avoid discussions on gender distinctions, not due to their unimportance, but because they deserve to be discussed in full in a separate study. Instead, I will focus on national origin, ownership of enslaved Africans, and religious affiliation, in order to highlight a spatial distinction between the east and west side of Gustavia.

Origins and population
According to the population summary sent to Stockholm in 1831, there were 2,460 inhabitants in Gustavia and 1748 in the countryside. 35 As can be seen in Table 1, the following years saw a steady exodus from the city of Gustavia, while the countryside had a gradual slight decrease in population. By the time the colony returned to French dominion, there were roughly twice as many people living outside as inside Gustavia. While there are no preserved nominal censuses of Gustavia from 1800 to 1815, when the colony arguably had its greatest influx of new residents from across the Atlantic world, the census of 1835 clearly shows that the population was for the most part not born in the colony, as shown in Figure 1. 36 Of the roughly 250 households, around 40 primary members were born in St. Barthélemy. Most of the primary members were born in the neighboring Leeward Islands, in particular the Dutch islands St. Eustatius and Saba, but also St. Kitts, Martinique, and Guadeloupe. Over 30 households had a primary member from St. Martin, although no specification was made whether this was on the Dutch or French side of the island. By contrast, there were approximately 30 primary members from Europe, with four of them from Sweden. Around eight primary members were born in Africa, of which five owned the lot where they lived.
There are certain trends of where primary member of different national origins resided, although these trends were often quite fluid. Swedes born in mainland Sweden often lived in the east part of Gustavia, the location of the government offices. The majority of the primary members from the British Caribbean lived in the west part of Gustavia, while more residents from the French Caribbean could be found in the eastern part. A section of lots near the harbor on the west part also had a large presence of primary members born in Europe, mainly France, but also England, Italy, Ireland, and Portugal.
During the 1854 census, there was a clearer homogeneity of origins. The number of primary members born in St. Barthélemy had increased from roughly 40-60, while groups of all other origins had declined. This was due to the exodus of the population of Gustavia, as fewer migrants arrived and previous residents of diverse origins either moved or died and were succeeded by their native-born children. The amount of property identified as uninhabited, whether due to being a church, a magazine, a garden, or a vacant lot, had increased from 208 in 1835 to 362 in 1854. One contributing factor to this was the fire of 1852, which occurred in the west, roughly in the same area where previously there had been a large number of European primary members. The number of lots with several primary members from different origins also doubled from 22 to 44. We can therefore assume that more households were now sharing properties, with a low level of immigration and high level of emigration. In the 1872 census, 96 of the 138 inhabited properties had people born in St. Barthélemy as primary members. 23 primary members were from the Dutch and British Caribbean, living primarily on the west side of Gustavia. By now, 390 properties were identified as uninhabited. As we have noted before, the rate of emigration from Gustavia was slower, but ongoing. The diversity of origins from different parts of Atlantic world initially make it hard to create a clear picture of Gustavia, other than that the plethora of nationalities would become more homogenous as the city's status as a free port became less conductive to immigration. However, we can see that there were some tendencies of nativity between the east and the west. The east part of the harbor was home to most Swedish officials, as well as people born in the French Caribbean. While those born in the Dutch Caribbean are seen across the city, as well as migrants from St. Martin, there were more people from the British Caribbean as well as Europeans on the west side. As a larger and larger share of the population was native to the island, it would seem that these divisions would have disappeared. However, if we examine this division from another angle, it becomes clear that differences between the east and the west remained.

Religious diversity or segregation?
By the 1860s, the categorization of religion in the censes emerged simultaneously as the categorization of race disappeared. Thus, when we examine the 1872 census, we can see the clearest division of Gustavia of the census categories. In Figure 2, it is clear that the east side of Gustavia was home to the majority of the Catholic population, while the west side featured the Protestant population, primarily Anglican, but also Methodist.
When we create a graphical representation of the census data, we do not have to rely on the religion of primary members but can create three categories: one in which all members of the household were Catholic, one where all were Protestant, and one where there were both Catholic and Protestant members within the same household. Here, it becomes increasingly clear that there is a division between Catholic and Protestant households, in part due to the fact that the south side sees a clear divide between the southeast and southwest.
So why was this? It was not the result of formation around the vicinity of churches, as the Lutheran, Anglican, and Catholic church were all in the south, within a few blocks from each other. The Methodist Chapel was located further west, but by no means far from the Catholic area. Furthermore, if we make a division between all Anglican and all Methodist households, Anglican households tend to be located further to the west, with no Anglican households on the east side. Moreover, the divide cannot be attributed to Protestants clustering around Swedish Lutheran officials, as the government offices lived on the Catholic east side. However, if we recreate a division from the 1835 census, based on whether or not the primary member comes from the Protestant or Catholic Caribbean, we can see a similar division, albeit with higher presence of Protestantism in the east. 37 We can also see that parts of the Catholic Caribbean represented by the Spanish Caribbean were all in the eastern Protestant area. The clearest way to recreate the religious division in 1872 is through the division between the primary member of the household being born in the British or French Caribbean, as seen in Figure 1. So, it is reasonable to assume that the religious division in Gustavia in 1872 was the reflection of a cultural division between those from the British and French Caribbean, as seen at least in 1835.
Furthermore, one could argue that the British Caribbean primary member deepest in French Caribbean territory was George Coxall from Grenada, an island which was heavily disputed between the French and British during the late eighteenth century. Likewise, the French Caribbean primary member deepest in the British Caribbean territory was Jane Shipton from Martinique, whose name suggests at least a cultural connection to the anglophone world. Thus, we can assume that they do not substantially challenge the assumption of a British and French Caribbean cultural divide across Gustavia. That being said, these results stand on their own, without highlighting these two households. The division between the east and west can be further nuanced if we examine the presence of slavery.
The Code Noir and the city St. Barthélemy was, like most of the Caribbean at this time, a slave society. This meant not only that slavery was legal, but that it was an integral part of the social structure of how society functioned and how it was divided. The Swedish colonial government had borrowed its legal structure from the French Code Noir, creating separate laws and punishments between whites, the enslaved, and free people of color. Free people of color were as numerous as whites in Gustavia and together they made up roughly half of the population of the city, the other half being the enslaved. The status of free people of color compared to whites is less clear in practice than in theory, as many held positions of wealth or prestige, despite discrimination. As a group, they were quite diverse, ranging from manumitted slaves to children of white merchants given a bourgeois education. 38 If we look at the 1835 census, most households did not have slaves. This is partially due to the fact that slavery as an institution in the Caribbean was beginning to diminish by this time, with the slave trade being outlawed by most European empires and slavery itself being abolished in 1833 in the British Empire. 39 As we examine Gustavia, shown in Figure 3, about 50-60 percent of all households had no slaves across the city. While there is a slightly higher number of households with slaves in general in the west, it is not until we examine households with six or more slaves that we notice a stark difference. While there were three households with more than six enslaved Africans on the east side and five on the south side, there were 16 on the west side.
When we examine the primary members of the households with more than ten slaves, as seen in Table 2, we can see that the majority are from St. Eustatius or Saba, islands from the Dutch Empire. Additionally, we can see that many have Anglo-Saxon names and a great many resided on the west side. There was no geographical division between a black and white neighborhood in Gustavia, as shown in Figure 4. There were some quarters where only people of color lived, free or enslaved, but the only commonality is that these were further from the harbor, and there were only two where there was more than one inhabited house according to the 1835 census -Berget and Dilly. This could be interpreted as higher property values for houses close to the harbor barring poorer free people of color from gaining access to harborside houses, but since there were also two quarters on the outskirts where no free people of color resided -Brinken and Engelbrecht -, it is more likely to be coincidental.  In her study of Basse-Terre and Pointe-à-Pitre, Pérotin-Dumon points out that the growth and expansion of the cities over 50 years coincides with the growth of the class of free people of color, demonstrating why most households owned by free people of color lived away from the coast, where most white households were. Thus, she indicates that zoning was not a factor, as much as socio-economic factors, as well as an unwillingness of free people of color to live by their former masters. 40 If we then compare this to Gustavia, which essentially grew to its full size in a ten-year period, we can identify urban growth as a primary factor. 41 Absent of any long-term expansion, no spatial divisions between whites and free people of color emerged. It could, however, also indicate that while unbound from slavery, many poorer free people of color still depended on proximity to white households for jobs in a service economy.
When examining the 1854 census, the decline in all white households is evident. By 1835, there were a total of 89 households with no free people of color, of which 39 had no enslaved people. By 1854, this had decreased to 13 households. Even with a sizeable reduction of the city's population, it is clear that the reduction in all-white households was much more significant. Additionally, all but one of the all-white properties were uninhabited by 1854, had completely new inhabitants, or had burnt down. In the quarter Krubban, white Engle Jeems lived with his white wife Rebecca, their eight children and white Joseph Garein of Puerto Rico. In 1854, his 43-year old daughter Caroline Jeems lived there with her three-year old daughter of color Ambrosine and 35-year old free man of color Louis A. Bralen. 42 This could be interpreted as a large emigration of white inhabitants from the city. The white population in Gustavia in 1854 was 56 percent of that in 1835, whereas the black population in 1854 was 64 percent of that in 1835, including the free and enslaved. However, as these statistics show, the decrease of whites was less dramatic than the decrease in houses without free people of color. While the white population overall decreased by almost half, the amount of all white households in 1854 was a third of what it was in 1835. If we break down this change into age groups, we can see that the decrease of whites was much more prevalent among those under the age of 15. Whereas there were 224 adult whites (15+) compared to 502 adults of color, roughly half as many, there were 85 white children (0-14) compared to 340 children of color.
What can we attribute this to? We could draw several conclusions from this data, either that integration between whites and blacks led to an increase in marriages and biracial children, registered in the census as children of color, or that white men abandoned their children, leading to a decrease of white adults simultaneous to an increase in children of color. The rapid decrease in all-white households indicates that this was not primarily a result of differing birth rates, as this would lead to a similar discrepancy among children born to all-white and other households. While there were four time as many children of color as white children, there were six times as many all-black households as allwhite, not counting the households of white and black people cohabiting.
We can also see the households in 1854 which contain former slaves emancipated in 1847there were 24 properties with one formerly enslaved inhabitant, 23 had 2-5, whereas four had 6-10. Half of the house with former slaves had no white inhabitants, while the other half had at least one. Often former slaves were employed as cooks and house servants, while others had independent jobs as sellers or laborers. Of the four properties with over six former slaves, only one had a white person present. This was property lot 418 in the quarter Godet, owned by carpenter John J. Aye, who was absent at the time of the census. Other than him, 17 people of color, of which six were former slaves, lived in four different households on the property. Similarly, 16 people of color lived in property lot 433 in the quarter Fyrfotan, 26 lived in the property 200 in the quarter Lyckan, and seven people of color, of which six were former slaves, lived on property 157 in the quarter Dälden. In all three of these cases, the owner of the property did not live there. When examining property lot 157, we can see that two of the women living there, Jeanne and Rose Auguste, were working as servants. What we can see is that work which previously would entail the enslaved living in the same household as their masters, could continue after liberation, with some living in the same household, while others instead lived independently.
By 1872, race was no longer a category in the census, which makes studies of racial differences in terms of space much more difficult. However, there are still denotations of those liberated from slavery in 1847, which by then made up 35 individuals. Of these, seven worked as washers, six as cooks, five as domestic servants and three as seamstresses. Others worked various jobs: nurse, cooper, huckster, carpenter, cigarmaker, while others were retired without working. There was no clear spatial pattern to where they lived. There were five households which had more than one emancipated person living there. Of these, two census entries suggest that they lived in their employer's household: Catherine,31,and Eleanor,63,living with Vice Fiscal T. W. Vaucrosson,and Sarah Dinzey,65,and Josephine Johnson,32, living with merchant R. Burton Dinzey.
What we can see from 1835 to 1872 is the gradual transition from a slave society to a post-emancipation society. In 1835 we can see slavery as a common, although not universal, aspect of the households of Gustavia. The west side had higher concentrations of enslaved people, but it is clear that they were present in the entire city. In 1854, directly after emancipation, we can see a decrease in the white population and all white households, most likely due to emigration. Labor relations with emancipated slaves working as servants continued, with some continuing to live in their employers' household, while other lived communally in properties with no to almost no white people. By 1872, while it is very unlikely that race did not factor into labor relations, it was no longer a census category, most likely reflecting its diminished role as an aspect for colonial organization and city management. Let us now return to what we have found out regarding the western area, as the socio-political dimension of this stratification becomes clear.

Political elites of western Gustavia
Like so many colonial documents, the censuses of Gustavia primarily reflect an administrative and gubernatorial perspective of the city, meant to illustrate often complex social bonds and spatial arrangements in the most efficient way possible. However, two patterns become apparent: First, a sociocultural divide between the eastern side of the harbor, home to several French-Caribbean Catholics from Martinique and Guadeloupe, as well as Swedish Lutherans, and the western side of the harbor, where a majority of British-Caribbean Anglicans from St. Kitts and Anguilla resided, as well as several Methodists. It is unclear to which degree this formation started as a religious division, or as a result of the rivalry between France and Great Britain, especially during the Napoleonic Wars during which the colony experienced its highest degrees of migration. Nevertheless, this trend can be seen as continuous from 1835 to 1872, even after the majority of Gustavia residents were born on the island. Previous research on the division between French and British subjects on the island has been sparse, with some indication of revolutionary parties being formed on the island and a French prize court being set up in the 1790s. However, there are no reports of actual conflicts between French and British factions. 43 Second, while the enslaved were present in at least half of the households of Gustavia, we can see that most of the households with a high number of enslaved peoplemore than six were located in the western Anglican side, a pattern that is even clearer for households with more than 10 enslaved people. Those born in the Dutch West Indies were most frequently among those owning a high number of slaves, followed by those from the British West Indies. The households that held the highest number of slaves within Gustavia were typically residents from nearby Dutch Caribbean islands with Anglo-Saxon names.
We can conclude that the western side of Gustavia was home to a Dutch and British merchant elite from Saba, St. Eustatius, and St. Kitts, who were overly representative in slave ownership. Their post-emancipation legacy can be seen in the presence of a religious division in 1872. However, the high amount of migration from the island means that the continuity in terms of property owners was very low. Many of these merchants cannot be seen after 1835, and the Protestant presence is a shadow of the Dutch and British merchants who left or died long before the 1870s.
There were also political aspects of this division because many of these residents cooperated within the political sphere. In 1810, there was a mutiny of the town militia of Gustavia, which led to the exile of several Swedish administrators. 44 Afterwards, a petition was sent to the Swedish crown prince by Jean Jacques Vaucrosson, Emmanuel Rey, John Bernard Elbers, John Joseph Cremony, and James Prince, all participants of the mutiny, explaining their reasoning. By examining their wills, we can see that all but Prince resided on the west side. 45 Similarly, we can look at Emmanuel John Stuart, who was charged with libel and treason for drafting a letter regarding the need to investigate a Portuguese ship in the harbor. From the trial we can see that Abraham Runnels read the letter during a visit, and that Cremony read it at a dinner in Stuart's home. He was later defended by the elected member of the council Peter Petersen, who would subsequently be accused of insubordination by the Governor. Stuart and Petersen both lived on the west side, as well as Abraham's son William. 46 This does not necessitate a shared political vision on all matters, however. While Stuart accused the Swedish government of not safeguarding against corsairs, Vaucrosson would host a party when British volunteers for Bolívar visited the harbor and other officers visited Cremony's estate in St. Martin. 47 Yet they were all part of a class of political and economic elites who predominantly resided in the same area, whereas the Swedish-born administrators lived across the harbor. Thus, a social space was created, where commercial bonds could be forged and attempts to influence colonial policies were planned and executed.
To bring this aspect into play for 1854, we can examine a petition sent to the King with 58 signatures from residents of St. Barthélemy. The petition concerned fees for passports, comparing the costs between the different Caribbean empires and how Swedish fees hampered competition and disincentivized trade from St. Barthélemy. It even questioned whether passports were necessary at all, citing the British Caribbean as one area where passports had been all but abolished, as well as the French, Dutch, and Danish Caribbean as islands with more lax restrictions. The petitioners also argued that after the abolishment of racial differences, the difference in payment for passports was also abolished. Now, everyone paid $1.08, the fee normally set for whites, thus making it harder for poor blacks to sell their straw hats on other islands. The timing of the petition was actually the recent death of the government secretary John August Uddenberg in 1852, as "sentiments of delicacy and consideration towards actual Incumbents of Office in this Island, have often had the effect of deterring the Inhabitants from making known, and of seeking redress of any Ordinance which has borne had upon them." 48 The petitioners feared that any complaint made under a current government secretary's rule would be judged as an attack on them personally, so a new government secretary would be necessary to judge the matter objectively.
As can be seen in Table 3, if we take the signatures of the petition and cross-reference them to the 1854 census, we can see that the west side becomes overrepresented in terms of signatures, with 11 signatures from the east, three from the south and 20 from the west, despite having equivalent population sizes -349, 311 and 446 persons respectively. The countryside had 19 signatures, while having a population of 1,691, making not just Gustavia, but west Gustavia in particular more representative of this petition. We can also see in Figure 5 that many signers in the west were also close neighbors. In would be an exaggeration to say that all signers were merchants or even wealthy. Several were carpenters, shoemakers, or farmers, while one signer was a four-year-old boy, who signed his name under his father's. However, I argue that this petition is still representative of a merchant class arguing for continued lowering of barriers for trade, often with particular concern for the political conflicts between them and the colonial administration. I also argue west Gustavia was the geographic center for this class, as they had been in the early 1800s. The signers of the petition could be aging merchants, steadfast in their free trade beliefs, their sons, eager to return the colony to its former glory, or their middle or lower-class neighbors, who were willing to sign the petitions of the businessmen in their immediate vicinity.

Conclusion
Gustavia is often described as an English-speaking town, in comparison to the Frenchspeaking countryside, which was often reflected in proclamations being issued to the town in English and the countryside in French. However, the English character has often been implied to be reflective of the town as a whole. I suggest that while English was most likely the lingua franca of Gustavia, the concentration of Anglo-Saxon religion and culture can be found among the slave-owning merchant elites of western Gustavia. These were connected to a larger Caribbean network and sometimes acted in alignment, but often in opposition to Swedish governance. In terms of residence, these two groups were quite literally on opposite sides.
If we place these findings within a larger historiography of the pre-democratic Caribbean, questions regarding culture and space arises. While Caribbean harbor towns, free ports in particular, have been shown to display amalgamations of various cultural influences from across the Atlantic World, the spatial dynamics between these have rarely been examined. 49 Despite harsh punitive divisions between members of racial categories that existed even after they were formally abolished, the clearest indicator of a spatial divide, at least in Gustavia, was not race, but national culture or religion. This could indicate that we underestimate the importance of these factors within nineteenth century Caribbean society.
It could also indicate that the necessity for free black labor prohibited clear spatial divisions, even within a relatively small city. Lack of segregation should of course not be assumed to be a sign of benevolence, but rather that spatial division between whites and free people of color was not a part of the toolbox of repression local administration utilized. As Marisa Fuentes has shown, while the repressive structure of slavery was often built into the architecture of the Caribbean city, it should not be seen as a segment of blocks reminiscent of North American racial segregation, a point previously made by Pérotin-Dumon and which this study emphasizes. Rather it was a maze, where homes and areas of comfort could be right next to symbols of oppression and exclusion.
Through GIS-analysis, we can unpack these cultural dynamics in ways that previously have been obscured, and demonstrate how they remained in Gustavia for decades. This form of GIS analysis could very well be employed in studies of other Caribbean cities, which could ask similar or other questions regarding social stratification, space, and early modern urbanization, especially in studying the social patterns before and after the abolition of slavery. Archival challenges will certainly require creative solutions and will require deviation from the technical methodology utilized in this study, just as this methodology was created within particular given archival constraints. Still, there is much to be gained from bigor mediumdata examinations of the Caribbean city.