Analysis of a novel source of data about early nineteenth-century Bombay, with a novel methodology, makes an important contribution to debates about inter-religious contact in South Asia. Bombay’s register of sales deeds, which lists the names of buyers, sellers, and neighbors, also permits identification of their religious affiliations when supplemented with archival information about the bureaucratic practices affecting property transactions. Findings suggest that property transactions within religious groups comprised most of the sales (60 percent). Contemporary petitions show that residents sometimes appealed to the state to prevent the sale of property to people who did not share their religion.

After the fire of 1803 in Bombay, landowners were asked to lease or sell their lands to people who would be evicted when the space around the town walls was cleared. Sunkersett Baboolsett, when asked to lease his oart or plantation at Girgaum, said that he would rent it “to any of the proper Gentoo cast[e]” (meaning to Hindu co-religionists), though he wished to retain at least some part of this land for “religious ceremonies.” Pillajee Nalljee said that he too would rent his lands only to those of his own “caste.” About the same time, the bishop of the Armenian Church in Bombay’s fort, expressed concern that his upstairs neighbor might sell the property to a Portuguese (Catholic) or Muslim buyer. Three years later, in 1805, Nillajee Madonjee, a Hindu resident, invoked his right of first refusal to prevent his Hindu neighbor from selling his plantation to build a Portuguese church (the Madonjee case, as it is called in the rest of this research note). A locality was described as “a part of the Town chiefly inhabited by Parsees,” suggesting the presence of a homogeneous neighborhood.

These diverse examples illustrate how religion and urban space intersected in early nineteenth-century colonial Bombay. Archival extracts, often involving petitions from residents, suggest that landowners attempted to control the religious composition of the groups that were renting their land or buying their neighbors’ land. Given the right of first refusal, and other instances of attempts to sell or rent property only to co-religionists, property transactions are ready terrain for analyzing inter-religious contact. This research note uses new methods and a register of property transactions to investigate cross-religious transactions and micro-spatial religious diversity.1

In the early nineteenth century, the East India Company ruled Bombay. A diverse religious, ethnic, and racial population lived inside and outside the town walls (the colonial fort). Evidence shows that commercial interactions and collaborations crossed religious and ethnic boundaries, and that different groups lived near each other. Cross-religious petitions and political action were frequent occurrences during this period. This study focuses on property transactions and urban space with the understanding that cross-religious contact in other domains could have taken different forms, remaining a question for future research.2

Attempts to further the scholarship about religion and cross-religious contact in colonial India must explore specific times and places to specify contexts and add new evidence rather than aim for transhistorical or pan-Indian generalization. Hence, this study takes three steps. First, it turns to the concept of segregation as an analytical device that bypasses some of the controversies in the literature about inter-religious contact in South Asia. Second, it relies on petitions and property registers that extend beyond colonial representations of religion and cross-religious contact. Recent literature about political practice in South Asia privileges such archival sources for their closeness to everyday life, although researchers tend to avoid them because they are more demanding to analyze and interpret. This study uses both a new archival data and a new methodology that is well suited to the South Asian context. Registers of sales deeds can illuminate the functioning of social categories, such as religion, since such documents were not created primarily for social categorization (as censuses were). Third, this study focuses on the early nineteenth century, which is considered crucial to the matter of inter-religious contact in India. Such contact prior to the later nineteenth century has attracted considerable debate.3

Segregation as a variable connotes the degree or extent of residential clustering by group; segregation can be either high (homogeneity) or low (heterogeneity or diversity). It is widely accepted as a quantitative measure, with established definitions and ongoing methodological innovations. The concept is used in various countries to understand racial, ethnic, religious, and economic segregation and the different mechanisms that drive them. Using segregation as a framing concept allows for a more exploratory analysis of inter-religious contact due to its flexibility as well as its attention to scale, a key conceptual category in the literature. This analysis focuses on the micro-spatial scale (immediate neighbors), as represented in the evidence.4

A data set of notarial documents, the register of sales deeds, has much to say about cross-religious social contact in the early nineteenth century. “Records of land sales are a treasure of information.… [They] served as title deeds and were likely to be copied and kept with great care for generations.” Property records are key aspects of social and economic history, detailing everyday market exchanges, information flows, petty land disputes, and relations between individuals and groups. Since the information contained in notarial documents is of a legal nature, and the interests of the transacting parties “contain an element of antagonism,” the information in such documents is likely verified and not one-sided. Although sales deeds often find use in quantitative analysis, this exercise is hardly straightforward. Interpreting such documents requires knowledge of bureaucratic land-registration procedures, the legal system of tenure, and customary laws or informal institutions relevant to property transactions. This study uses this data set of property transactions for the first time to conduct a micro-spatial investigation into residential diversity by religion in the early nineteenth century.5

The material within the archival sources yields three types of information: (1) the practices of segregation, such as right of first refusal, and the layout of streets and houses; (2) the procedures of land registration; and (3) a quantitative analysis of the register of sales deeds. Information about the practices of segregation and the procedures of land registration provide context for the quantitative measures of segregation. The records of this period, located in the Maharashtra State Archives, Mumbai (msa), are unpublished and handwritten.

Information about segregation practices throws light on the concrete mechanisms and actions by which religious groups managed contact with out-group members in colonial Bombay. The Building Committee and Town Committee diaries from the late eighteenth and early nineteenth century contain policies and plans, internal state memos and correspondence, and petitions from residents that show the strategies of different groups to maintain social distance. The diaries also provide information about the layout of streets and houses that permit inferences about the social significance of the physical distance between people and groups. Descriptions of the bureaucratic procedures of land registration in Bombay—primarily Hall’s Remarks of 1803 and Warden’s Report of 1814—serve to evaluate the register of sales deeds as a data source. The third source of evidence, the register of sales deeds, enables discoveries about cross-religious property transactions and the religious diversity of micro-neighborhoods.6

The primary data for this research note are sixty-four sales deeds in a register from 1802 to 1804 (60 percent of which are from 1803); the only transaction excluded from the analysis was a mortgage. The sales comprise 76 buyers, 90 sellers, and 167 enumerated neighbors, all of whose names were coded for religious affiliation, as explained below. The two sets of information—about buyers and sellers and about neighbors—illustrate two distinct aspects of social contact between people who were not co-religionists. The names of buyers and sellers reflect their religious identity and provide an indication of property transactions between religious groups. The names of immediate neighbors enables an exploration of religious segregation/diversity at a micro-spatial scale (hence the term micro-neighborhoods).7

The material about the bureaucratic procedures of property registration reveals the value of this register as a source of data. The colonial state in Bombay began to register property sales and transfers by at least the early 1700s (Warden mentions a date as early as 1715). Contemporaries’ evaluations of the procedures are mixed. Reflecting on the history of land transactions, Warden pointed to the government’s pursuit of “a lax system, in leasing out the public property.” Warden (1814) had a negative evaluation of the land-registration bureaucracy, but Hall (1803) was more positive, asserting that “Houses were conveyed by regular Deeds of Sale … and the conveyances duly registered.” Warden noted that regardless of the ambiguity of the tenure system existing at the time, and reflected in this register, residents were buying and selling property with the belief that they were engaging in a transfer of property in perpetuity, just as people do today. The state had a strong incentive for registering sales deeds because they were the basis of rent collection.8

The evidence does not suggest any systematic tendencies in the registration of property sales (such as a strong trend toward registering transactions between people who were not co-religionists). Given the lack of prior theorizing about selection bias for this type of data, the degree to which the transactions in this register are representative of the population of property transactions or of micro-neighborhoods is difficult to ascertain (more below). This research note offers suggestive information about cross-religious property sales and micro-spatial diversity; it does not allow conclusions about religious segregation at the city-level.9

Each handwritten sales deed of a house or oart (plantation) bears the signature of James A. Grant, the Secretary to the Governor (a high-ranking colonial bureaucrat). The actual entries would have been created by British or Indian clerks or scribes. For houses, the deeds usually specified whether the property was inside or outside the town walls, denoting the spatial importance of the fortifications in the land market. Land values were higher inside the town walls; people were willing to pay a premium for easy access to commercial transactions and for the security that the fortifications provided. The Collector’s statement that land values within the walls were “many hundredfold” higher than values outside the walls was certainly an exaggeration, but it strikes the right note.10

Oarts, which were mainly located outside the fort, are a now-forgotten part of Bombay’s topography. At one time an important type of settlement in the city’s early history, they slowly succumbed to urban development and government acquisition. Oarts, which often carried names, could be owned partly or fully. Their value lay especially in such fruit-bearing trees as coconut or jackfruit. The houses located on some oarts were available for rent. In the Madonjee case mentioned earlier, the oart sold by the neighbor putatively comprised “thirteen houses large and small … occupied by thirteen Hindoo families containing between fifty and sixty people.” Oarts encompassed a much larger plot of land than did houses. Houses were located on smaller plots usually within the town walls, closer to neighbors than were oarts and therefore, in all probability, more religiously segregated.

The deeds recorded the name(s) of buyer(s) and seller(s) and the amount of the transaction in rupees, noting the date of both the sale and registration and sometimes other pieces of information. The deeds, which were usually, though not always, witnessed by one to five people, generally provided the names of neighbors in all four directions as a way to locate plots and to delineate their boundaries. The plot measurements appeared in varying units, predominately feet or yards though occasionally in more arcane metrics like “covits,” “sticks” or “catties.” The metric “wheel” found use for oarts. The text of a deed registered on September 23, 1802, is a prime example: “Lalla Hemchund Banksally … [do] sell, release and confirm unto the said Hormuzjee Dadysett his heirs and assigns all that lower roomed House and appurtenances which house he had as a gift from Rattana [Naiquia] of [Meed] … and covered with cadjan situated in the woods in the oart called Saint Antonio belonging to Govindjee Crustnajee Weaver of Bombay and bounded to the Eastward by the House of late […] and John Harris, to the North Hormuzjee Dadysett the Purchaser, to the Southward by the House of Candoo Darmajee and a water well and Westward a small passage and a House of Nagaindass.”11

The deeds for the houses to be sold sometimes described their construction and amenities (tiled roofs, stone walls, wells, etc.). The house in the extract above was covered in cadjan (thatch made of palm leaves); even many wealthy houses had thatched roofs. Lower rooms (on the ground floor) were less desirable because of the negative effects of weather, dampness, and water seepage. Deeds for houses sometimes mention passages—the narrow alleyways between buildings to which customary rights were attached. These passages were a common source of nuisance complaints or lawsuits in the files of the Committee of Buildings, since they could be used for water run-off and sanitation. Neighbors, whether co-religionists or not, who shared such liminal spaces were more likely to enter into conflict about them.12

The list of the names in the property transaction above is illustrative. The oart, which carried a Portuguese name (Saint Antonio), belonged to Govindjee Crustnajee (or, in the present-day spelling, Govindji Krishnaji), both first and last names being of Hindu gods. The seller of the house in the transaction was Lalla Hemchund Banksally (Hindu) and the buyer Hormuzjee Dadysett (Parsi). The neighbors included the late John Harris (European), the Parsi purchaser (Dadysett), and two Hindus (Candoo Darmajee and Nagaindass). The suffix jee (present-day ji) marked Parsi and Hindu last names. “Darmajee,” a Hindu name, derives from darma or dharma, a Hindu religious concept unlikely to serve as a name in other religions. “Nagaindass” appears in the list of Hindu names in the 1884 Bombay Directory (as Nagindass, -dass or -das being a common suffix in Hindu names). This classification of names indicates that the transaction for this particular property cut across religious boundaries, and that its micro-spatial neighborhood was diverse.13

The focus herein is on the four main religious groups—Christian, Hindu, Muslim, and Parsi. Internal divisions between these groups were also consequential; different sects, castes, ethnic groups, or tribes existed within all the main religious groups (although our methodology does not allow for an analysis of intra-religious differences, they could have been important in spatial demography and segregation practices). In the same vein, this research note refers to the Portuguese, Indo-Portuguese, and Indian Christian residents collectively as Christian; the classification of names does not allow us to distinguish between these groups.

Did these overarching religious categorizations have social meaning during this period? Everyday documents such as petitions and sales deeds can help to answer this question. Drawing from such evidence helps to avoid essentialist or anachronistic interpretations of ascriptive categories. Instead, it enables us to explore the extent to which these categories were active, whether other descriptive choices were available at the time, and how such categories might have structured social interaction. In the Madonjee case, the petitioner refers to himself as a “Hindu” and to the Portuguese as “Portuguese,” rather than invoking ethnic, regional, or caste sub-groups of the heterogeneous “Hindu” or “Portuguese” category.

Were such religious categories imposed on these documents by scribes or translators? The category of Hindu had wide use during this period in colonial accounts, court documents, travelogues, lists, and self-identifications in legal documents. Since Hindu sub-groups, such as castes, also appear in various colonial documents in the early nineteenth century, however, the implication is that the appearance of overarching religious categories was not merely a result of poor translation or the unavailability of intra-religious categories. In other words, when Madonjee spoke of “Hindoos,” he was making a discursive choice among the many sub-categories available to him as well as to scribes. Similarly, the overarching category of Parsee/Parsi was common, except in intra-religious disputes when finer categorizations could be significant.14

Information about the religious affiliation of buyers and sellers and of immediate neighbors forms the basis of two analyses herein—an analysis of the frequency of cross-religious property sales in the property register and an analysis of the diversity of micro-neighborhoods, or the spatial contiguity of people who were not co-religionists. The second analysis includes a binary coding of whether a micro-neighborhood was homogeneous; a re-classification of neighborhoods to include the religion of the seller (current neighborhood); and a redoing of the binary coding to include the religion of the buyer (prospective neighborhood).

Names often carry information about group identity, including race, religion, caste, ethnicity, or region. The classification of Indian names by religion or caste in the recent literature involves the development of algorithms from lists of names already classified by group identity (for example, in the census) and using them to classify other lists. This note follows the logic of this classification method, though not the algorithmic procedure. Drawing from a handwritten data set compiled in a period prior to lists or standardized spellings meant that each name had to be coded manually rather than via algorithms. The lists of names classified by religion in the Times of India Calendar and Directory for Bombay (1821 and 1884) and in other archival sources were cross-referenced with names from the register of sales deeds.15

First names proved more useful than last names (Mohamed, Ganesh, Antonio, et al.). Additional information often listed with the name of the buyer and seller were, for example, caste references (Bramin and Battia, Hindu caste names) and religious affiliation (Parsey, Moreman, Bhori, Coja, et al.). Names such as “Cavasjee Jivanjee Parsey” or “Bhugwan Ressow Woopathai Bramin” can be ascribed to Parsi and Hindu groups, respectively (Parsey=Parsi, Bramin=Brahmin caste of the Hindu religion). This listing of religious or caste affiliations in sales deeds (but also in building permits, petitions, lawsuits, and trial testimonies) is an important aspect of social history, illustrating how persons carried their religious and caste identity into the public domain. It also confirms that the analysis of names can provide insights into group identity and inter-group or intra-group interaction.16

The register of sales deeds, together with other archival information, sheds light on the social history of property transactions in early nineteenth-century Bombay. Property could be sold at public auction, sold by the sheriff, sold as part of an estate, or sold through attorneys and brokers. Sales were often advertised in the Bombay Courier, through public notices (including the beating of a drum by a town crier). Sales required some form of public notice and a twenty-one-day waiting period to register any disapprovals or competing claims, presumably including the right of first refusal. In 1806, an ad in the Bombay Courier announced the sale of a property after the death of a landowner, “by Public Outcry at four o’clock in the afternoon on the Premises a piece of Ground situate in Borah Street, within the Town Walls, the property of Mussooboy wife of Bapoo Vissoo Sinoy Wagla late of Bombay Gentoo.”17

This register listed forty-three houses of the sixty-four properties transacted. The other twenty-one were oarts (or plantations), which themselves could have included houses, though this study does not code them for a house sale without explicit mention of a house. Six houses lay within the town walls, six in oarts, and twenty-seven outside the fort or town walls but not in oarts—likely in what was called a “native town” or “black town.” Four houses were impossible to classify as lying inside or outside the fort.

The average price of houses transacted was 2,639 rupees. The average price of the thirty-two houses outside the fort was 1,866 rupees, and the average price for the six houses inside the fort was higher, as expected, 7,175 rupees. The average price of the twenty-one oarts that were sold without any house listed in the transaction was 2,093 rupees. The deed for the least expensive property, which cost 115 rupees, did not include any details that can explain such a low price. The most expensive property was a large plantation with hundreds of fruit trees, a bungalow house, and sheds, worth a total of 12,251 rupees.

Women as Sellers

In eighteen cases, the sellers were one or more women (28 percent), but in only one case was the buyer a woman. These figures are important, since not much is known otherwise about women as owners or transactors of property in colonial India. The register most often identified women as widows but sometimes as sisters, mothers, or daughters, or just as “woman,” generally with a first name only—for example, “Parvatty Gardner Woman” or “Putlaboy widow to the late Wesswanathjee Putlajee.” Although inheritance laws and customs varied across religious groups, widows appeared as named transactors in the selling of property across all four religious groups. The fact that many more women were sellers than buyers implies that women came into property through inheritance rather than through market transactions. The large number of women sellers who were widows suggests that many women entered the land market through the death of a male property owner. The archival record does not establish whether women sellers were selling property of their own accord, nor how long after coming into ownership were they selling it. Inheritance was largely due to widowhood, although scattered cases of siblings selling property also appear in the sales register. For example, “We Hirjee Tucajee Coppersmith, Jumnaboy and Annundeeboy, women, Brother and Sisters, jointly and separately have sold … an House situated without the Town Wall in the Poydowney.”18

The sixty-four sales transactions, with a total of 128 transacting parties, comprised seventy-six buyers and ninety sellers. All but seven of the transacting parties could be coded with certainty for religious affiliation. The religious affiliation of five of the uncertain parties, however, could be estimated with some confidence, leaving only two names completely unclassifiable. A secure 97 percent of the transactions qualified for the analysis.

Table 1 is a cross-tabulation of the religious affiliation of buyers against sellers; the columns represent the religion of the sellers, and the rows represent the religion of the buyers (Hindu, Muslim, Parsi, Christian, and unclassified). Figures on the off-diagonal represent sales to those who were not co-religionists, that is, cross-religious sales. Figures on the diagonal of the table represent sales to co-religionists—Muslims to other Muslims, Christians to other Christians, etc.. Findings show that transactions reflecting intra-religious transactions represent 61 percent of all the transactions in the data set (thirty-eight of sixty-two, excluding the two unclassified transactions); the 44 percent (twenty-seven) between Hindus comprised the largest group of intra-religious transactions.19

Table 1

Religious Affiliation of Buyers and Sellers

Hindu 27 36 
Muslim 11 
Parsi 10 
Total 39 12 64 
Hindu 27 36 
Muslim 11 
Parsi 10 
Total 39 12 64 

note Pearson chi2(12)=32.4959; Pr=0.001.

No existing benchmarks dictate how to interpret these numbers. Is the 61 percent proportion of the total transactions in the data set high or low? The statistically significant difference in the distribution of sales across buyers and sellers of different religious groups indicates that sales were not randomly distributed across religious groups. The tendency to transact property with co-religionists was greater than that with those who were not co-religionists, although property sales across religious divides occurred as well. This finding also throws some light on whether social categories such as Muslim or Portuguese had social relevance during this period.20

Intra-religious Transactions

As mentioned earlier, overarching categories such as Hindu, Parsi, or Muslim embody various heterogeneities and intra-group differences (caste, ethnicity, tribe, region, etc.). Table 1, which is based on broad religious groupings, shows a tendency toward intra-religious transactions, but it cannot say whether a particular sub-group (say, a Parsi sub-group) was more likely to transact with its own sub-group rather than with any Parsi buyer or seller. The figures in Table 1 implicitly include the intra-religious transaction data that we would see if we could classify sub-groups. If religious sub-groups were more likely to transact with those who were not co-religionists, Table 1 would not show a significant tendency for intra-religious transaction across the four main religious groups. In other words, the four main religious groups do not impose structure on the data. The categorization of names does not permit any findings beyond the main groups, but the higher tendency for intra-religious transactions, along with the religious categories advanced in the petitions, suggests that the four broad religious groups had genuine social meaning.21

Another way to quantify this relationship is through odds ratios: Computing the tendency of Hindu sellers to sell to Hindu buyers rather than to other groups, relative to the tendency of non-Hindu sellers to sell to Hindu buyers rather than among themselves, obtains an odds ratio of 5. Hindu sellers were five times more likely to sell to Hindu buyers than to others, as compared to the likelihood of non-Hindu sellers to sell to Hindu buyers instead of among themselves. This odds ratio suggests a tendency to sell to co-religionists among Hindus. Odds ratios can be similarly computed for any other group of interest.22

Intra-religious transactions on the diagonal need not necessarily signify trust, or preferences for neighborhood contiguity; co-religionists could have reasons to maintain social distance (for example, caste or sectarian differences). Theoretically, some might even have preferred to buy from those who were not co-religionists since such transactions were more likely to be disinterested, arms-length, and market-oriented. Nor are the cross-religious property transactions on the off-diagonal necessarily indicative of commercial collaboration since properties were often sold at auction or through brokers without face-to-face interaction. These cross-religious transactions are not the same as co-owning a business concern, serving as a lawyer, or maintaining a long-term professional relationship with someone who was not a co-religionist.23

We might suspect that those transacting property with people who were not co-religionists would register a transaction more promptly because of less social trust. The opposite, however, was the case. The time between the date of sale and date of registration for properties transacted between people who did not share religion was longer than it was for properties transacted between co-religionists (428 days versus 139 days, statistically significant at p=0.04). The time between the date of sale and the date of registration can serve as a proxy for a perceived need for complete documentation of a property transfer. This metric therefore helps to analyze selection bias, which is a critical methodological question in any attempt at quantification from the archives. This finding implies the absence of a greater tendency to register properties transacted with out-group members, or people identifying with other religions.24

This part of the analysis is based on the names of buyers and sellers of property; it does not include the names of tenants or sub-tenants. Considering the question of class, at least some of these buyers/sellers were members of the elite. Some of the individuals in the property transaction register were well-known at the time. For example, Ballajee Shamsett, who bought a house within the town walls in 1803, was a wealthy Hindu goldsmith who owned several houses in a street named after him. Similarly, Hormuzjee Dadysett, the purchaser in the extract presented earlier, belonged to a well-known wealthy Parsi family. Even in the case of elite transactions bought as investments—particularly oarts—neighbors could resist selling to those who were not co-religionists (as in the Madonjee case). Many owners of oarts did not live on site. They rented the plantations to others or farmed the land and sold the houses to others. For example, a few years after this period, the revenue surveyor remarked that “few of the … [landowners were] living at, or near their oarts.” Nonetheless, the Madonjee case involved a landowner protesting against the sale of a neighboring oart to someone outside his religion, basing his appeal on the desires of the families in the neighboring oart to live near co-religionists. Similarly, non-elite individuals could also resist contiguity by objecting to the sale of neighboring property to those who did not share their religion.25

Most of the registered deeds enumerated the immediate neighbors of the transacted property in four directions (88 percent of deeds enumerated at least one neighbor). The property under transaction can be seen as the center of an ego network in a micro-neighborhood. In principle, the seller was part of the current micro-neighborhood, and the buyer was part of the prospective micro-neighborhood. When the name of one of the neighbors is unclassifiable, the case need not be disqualified from this analysis if the names of neighbors that can be classified indicate heterogeneity. We cannot say with full confidence that a neighborhood with two Muslim names and a third that is unclassifiable was homogeneous or not. But a micro-neighborhood with two Muslim names, one Christian name, and one unclassifiable name can be classified as heterogeneous/diverse even without the information from the unclassifiable name. The presence of one unclassified neighbor does not change the heterogeneity code of a neighborhood with classifiable heterogeneous neighbors. Among the names of the 167 neighbors, fifteen were unclassifiable, or 9 percent. Four names classified with less than full confidence are also included in the analysis. The presence of a single European name among the immediate neighbors is not a surprise; at this time, Europeans concentrated inside the fortified city, at a physical distance from the Indian population.

The measures outlined earlier reflect the degree of spatial contiguity between different religious groups. Table 2 shows the percentage of homogeneous neighborhoods based on these three measures, separately for houses and oarts. In the first column (all properties), 48 percent of the micro-neighborhoods enumerated in the register were homogeneous based on the first binary measure that includes immediate neighbors but excludes the religious affiliation of buyers and sellers. Thus, more than half of the neighborhoods had at least one immediate neighbor who was not a co-religionist. In contrast, 41 percent of micro-neighborhoods were homogeneous when the religion of sellers was taken into account, and only 30 percent when the religion of buyers was included. Moreover, neighborhoods around houses were less diverse than those around oarts—almost 70 percent of these neighborhoods being homogeneous compared to just 20 percent of those around oarts, based on the first measure. The analysis draws only from enumerated neighbors around transacted properties, which is not a random sample of all properties. This analysis is therefore suggestive; it does not allow conclusions about city-level segregation patterns.

Table 2

Religious Segregation in Micro-Neighborhoods

 % homogeneous all properties% homogeneous in houses% homogenous in oarts
Micro-neighborhood 48% 69% 20% 
Micro-neighborhood including sellers 41% 63% 11% 
Micro-neighborhood including buyers 30% 44% 10% 
 % homogeneous all properties% homogeneous in houses% homogenous in oarts
Micro-neighborhood 48% 69% 20% 
Micro-neighborhood including sellers 41% 63% 11% 
Micro-neighborhood including buyers 30% 44% 10% 

note The column “homogeneous in houses” excludes houses in oarts.

There are two patterns evident in Table 2. First, micro-neighborhoods are least homogeneous (that is, most diverse) when they take into account the religious affiliation of buyers. Second, oarts were significantly more diverse than houses, in line with prior expectations. Oarts were more likely to be bought as an investment and to lie at a greater physical distance from each other than were houses, thus reducing the likelihood of contact with neighbors, as well as any impact of contiguity with those who were not co-religionists. In other words, the analysis of micro-neighborhoods demonstrates the interrelationship between social and spatial distance; the greater was the spatial distance, the more diverse were the micro-neighborhoods.26

Residents bought property not only as an investment or to rent, but also as homes for families or co-religionists, as charitable contributions, or as sites for shops and workshops. The occupations of some enumerated neighbors suggest that the residents were not wealthy speculators but non-elite workers or craftsmen (oilmakers, potmakers, gardeners, and carpenters). Thus, the first measure of homogeneity (which excludes buyers and sellers) might be the best measure, more accurately enumerating those who actually lived on the property rather than those who may have owned the land but did not occupy it. Although the data about buyers and sellers in the foregoing analysis includes at least some elites and non-resident landlords, the data about immediate neighbors are more likely to reflect owner occupiers and tenants.

How would people have experienced religious diversity or segregation at the street level in early nineteenth century Bombay? Recent research on racial segregation stresses the importance of the street or street segment for the study of segregation. Qualitative evidence about segregation practices and the layout of streets and houses can help us to understand social distance and the likelihood of interaction between those of different religious persuasions. From the late eighteenth century onward, “encroachments”—architectural projections, balconies, extensions, benches, and steps that protruded into public streets of the town (in the non-oart areas)—were a concern of the Company’s government. These extensions suggest a narrowing of already narrow streets, and a busy street life. For example, a petition from residents within the town walls referred to the “universal custom for the shops to have benches before them.” Many of the properties transacted in this data set bordered on public streets where neighbors routinely ran into each other and where private space might encroach into the public arena. As mentioned earlier, the sharing of liminal areas, like the passages between buildings, resulted in problems with sanitation, shared walls, and water run-off, but they also provided spaces for people to socialize, such as benches. Given what we know about the design of houses and extensions at this time, neighbors, not just co-religionists, had many opportunities to interact with each other.27



This research note contributes to the literature about inter-religious contact, focusing on a period that has not seen much scrutiny. Diverse archival sources suggest that religion and urban space intersected in early nineteenth-century colonial Bombay. Evidence from petitions indicates that some residents expressed a preference to avoid contact with people outside their religion, and/or that this preference could be used to justify an attempt to control urban space. The analysis herein of a hitherto unexplored data set of property transactions, and original historical research about bureaucratic procedures of property registration, finds that 61 percent of transactions were to co-religionists, and the rest crossed religious boundaries. Micro-neighborhoods could be diverse, although the degree of diversity varied by the measure of diversity and by the settlement type; micro-neighborhoods around houses were significantly less diverse than those around oarts.

A significant finding that emerges from this research is the distinction between houses and oarts as forms of property. Oarts or plantations were a significant part of Bombay’s early settlement patterns and political economy. Many of these large estates functioned as both productive and residential spaces. Because oarts were more likely to have been bought as investments rather than as residences for their owners, their neighborhoods tended to be more diverse, and the relationship between the religions of buyers and sellers to be much looser, than in neighborhoods around houses.

The use of data from notarial documents comes with caveats. Although intra-religious cleavages are likely to have been consequential, the method used herein permits an analysis of only the four main religious groups in Bombay. Similarly, evidence about the representativeness of the sales register is ambiguous; some contemporaries argued that the registration bureaucracy was “lax,” whereas others saw it as efficient. This study innovates a measure of selection bias based on the time lag between the date of sale and the date of registration. Selection bias would exist in the case of a greater propensity to register sales transactions with those who were not co-religionists (such transactions would be registered closer to the date of sale than would intra-religious transactions). The absence of evidence for this type of selection bias provides some confidence in the register as a source of data.

The preceding analysis provides pioneering information about cross-religious property transactions and the diversity of micro-neighborhoods in early nineteenth-century Bombay. As the nineteenth century progressed, however, inter-religious contact changed as population and urbanization increased, and local and national politics evolved. The history of these changes and their impact on urban segregation patterns remains to be written.

The methodology for the use of property registers as sources of data, and the methodology for the analysis of names, has little, if any, precedent; the painstaking archival work of transcribing and digitizing such an arcane data set is always a gamble. As more records become available in digital format and thus more readily searchable, names and other information relevant to social history can become a fruitful area of study. Future research could focus on other unusual sources of data from the early nineteenth century or even earlier. New sources of empirical evidence can bring cross-religious contact into better relief.


In this period, caste was used interchangeably with religion; it did not refer only to Hindu caste groups. Town Committee Minutes, 11/19/1803, Town Committee Diary 1/183, Maharashtra State Archives, Mumbai, India (hereinafter, msa); Petition of Jacobus, Armenian Archbishop (not dated, probably 1802), Committee of Buildings Diary 3/179, 1799–1803, msa. For another case of residents attempting to prevent property sales citing religious grounds, see Coppersmiths’ petition against Parsi resident, Committee of Buildings Diary, 4/180, 1803–07, msa. The plot of land in the Madonjee case was in the Esplanade outside the town walls. Letter from Nillajee Madonjee and others, 1/25/1805, to Governor Duncan, Town Committee Diary 3/185, 1805, msa. Right of first refusal or preemption as studied in the agrarian context and in pre-colonial South Asia suggests that this practice did not emerge solely from colonial ideas about religion and spatial separation. See Dharma Kumar, “Private Property in Asia? The Case of Medieval South India,” Comparative Studies in Society and History, XXVII (1985), 340–366; David Washbrook, “Law, State and Agrarian Society in Colonial India,” Modern Asian Studies, XV (1981), 649–721. For examples from urban South Asia, see Mariam Dossal, Theatre of Conflict, City of Hope: Mumbai 1660 to Present Times (New York, 2010), 58–59; for broader questions about the intersection of religion and caste in urban space see Kanakalatha Mukund, “Caste Conflict in South India in Early Colonial Port Cities, 1650–1800,” Studies in History, XI (1995), 1–27; Patrick Roche, “Caste and the British Merchant Government in Madras, 1639–1749,” Indian Economic and Social History Review, XII (1975), 438–407. Committee of Buildings diary, 1/177, 1787–1793, msa.


The closest population figures from 1813 indicate 250 English people, 5,464 Parsis, 4061 Hindus, 775 Moors, 146 Portuguese, and 105 Armenians within the town walls. The total population of Bombay up to the limits of Parel was 140,000. Dinshaw E. Wacha, Shells for the Sands of Bombay Being My Recollections and Reminiscences (Bombay, 1910). For literature on the colonial city, the dual-city model of the racially segregated colonial city, and its recent revisions, see William C. Bissell, “Between Fixity and Fantasy: Assessing the Spatial Impact of Colonial Urban Dualism,” Journal of Urban History, XXXVII (2011), 208–229; Carl H. Nightingale, “Before Race Mattered: Geographies of the Color Line in Early Colonial Madras and New York,” American Historical Review, CXIII (2008), 48–71. Selected histories of Bombay include Mariam Dossal, Imperial Designs and Indian Realities: The Planning of Bombay City, 1845–1875 (Bombay, 1991); Robert Lewis and Richard Harris, “Segregation and the Social Relations of Place, Bombay, 1890–1910,” South Asia: Journal of South Asian Studies, XXXVI (2013), 589–607; Meera Kosambi, Bombay in Transition: The Growth and Social Ecology of a Colonial City, 1880–1980 (Stockholm, 1981). Petition from cultivators of salt batty ground, 24/3/1805 and 13/6/1805, Town Committee Diary, 3/185, msa. This petition included Hindu and (Indo-)Portuguese cultivators. For how inter-religious conflict in some domains could co-exist with consensus in others, see Christopher A. Bayly, “The Pre-History of “Communalism”? Religious Conflict in India, 1700–1860,” Modern Asian Studies, XIX (1985), 177–203.


In the words of Jeremy Menchik, “Review: The Constructivist Approach to Religion and World Politics,” Comparative Politics, XLIX (2017), religion, “like other aspects of culture and identity, is heterogeneous over time and space, multifaceted in practice, and its relevance to politics is dependent on context” (562). Research on inter-religious contact and conflict in South Asia is often filtered through the literature on “communalism.” For the key debate, compare Gyanendra Pandey, The Construction of Communalism in Colonial North India (New York, 1990) to Bayly “Pre-history of Communalism.” For critiques, see Ayesha Jalal, “Secularists, Subalterns and the Stigma of ‘Communalism’: Partition Historiography Revisited,” Modern Asian Studies, XXX (1996), 681–689; Peter Heehs, “Indian Communalism: A Survey of Historical and Social-Scientific Approaches,” South Asia: Journal of South Asian Studies, XX (1997), 99–113. Aparna Balachandran, “Petitions, the City, and the Early Colonial State in South India,” Modern Asian Studies, LIII (2019), 152; Nita Kumar, review of Gyanendra Pandey, The Construction of Communalism in Colonial North India (Delhi, 1990), in The Indian Economic and Social History Review, XXIX (1992), 230; Sandria B. Freitag, Collective Action and Community: Public Arenas and the Emergence of Communalism in North India (Berkeley, 1989); Veena Das (ed.), Mirrors of Violence: Communities, Riots and Survivors in South Asia (Delhi, 1990); for the challenges of working with property and tax data to study historical segregation, Clé Lesger and Marco H. D. Van Leeuwen, “Residential Segregation from the Sixteenth to the Nineteenth Century: Evidence from the Netherlands,” Journal of Interdisciplinary History, XLII (2012), 333–369. Much of the controversy in the communalism literature concerns the nature of contact in the pre-1860 period (the prevalence of conflict and the influence of colonialism). See Surya P. Upadhyay and Rowena Robinson, “Revisiting Communalism and Fundamentalism in India,” Economic and Political Weekly, XLVII (2012), 35–57.


Angelina Grigoryeva and Martin Ruef, “The Historical Demography of Racial Segregation,” American Sociological Review, LXXX (2015), 814–842; John R. Logan and Benjamin Bellman, “Before The Philadelphia Negro: Residential Segregation in a Nineteenth-Century Northern City,” Social Science History, LX (2016), 683–706. Logan, “Making a Place for Space: Spatial Thinking in Social Sciences,” Annual Review of Sociology, XXXVIII (2012), 507–524.


Aglaia Kasdagli, “Notarial Documents as a Source for Agrarian History,” Hesperia Supplements, XL (2007), 59, 54. Research on segregation in late nineteenth-century Bombay includes Jim Masselos, “Social Segregation and Crowd Cohesion: Reflections around Some Preliminary Data from 19th Century Bombay City,” Contributions to Indian Sociology, XIII (1979), 145–167; Lewis and Harris, “Segregation and Social Relations.”


Warden, “Report on Landed Tenures”; Hall, “Remarks,” Selection 55A, 1803, msa. Francis Warden was Chief Secretary to the Government in 1814. Phineas Hall was an attorney in the East India Company, Bombay 1803.


Register of Deeds of Property, Selections 34, 1801–06, msa. Since only a few of the addresses are identified by street name and number or the name of the oart, establishing geographical location is difficult, except for whether a property was within the fort.


Dossal, “Theatre of Conflict”; Warden, “Report on Landed Tenures,” 111, msa; Hall, “Remarks,” 42, msa. Other registries relevant to rent collection include the Collector’s Books, which are referenced often in the archival record (for instance, in the Town Committee Diary 1/183), but they are missing from the archives. Such revenue-related documents could be in the state revenue office to which scholars have only limited access.


Jamaican colonists would not have been likely to be “systematically indifferent” about registering property titles either. Ahmed Reid and David B. Ryden, “Sugar, Land Markets and the Williams Thesis: Evidence from Jamaica’s Property Sales, 1750–1810,” Slavery & Abolition, XXXIV (2013), 403.


The comparison was made to areas outside the walls such as Byculla, dated 8/5/1800, 376, Town Committee Diary, 3/185, 1805, msa. Similarly, those buying houses within the town walls in the years preceding 1814 “paid more for them in consequence of their situation, then they would otherwise have done” (“situation” implying location; quoting the company’s counsel in Warden, “Report on Landed Tenures,” 73).


A catty was ten feet and a covit a yard. A similar transposition for the metric wheel is difficult to find.


Chattopadhyay, “Blurring Boundaries”; Vanessa Harding, “Space, Property, and Propriety in Urban England,” Journal of Interdisciplinary History, XXXII (2002), 549–569.


The Parsis are a religious group worshipping Zoroaster who migrated to India from Persia (Iran). They created a wealthy and commercially significant community in colonial Bombay.


The question of when pan-Indian religious identities emerged, and the extent to which they were “constructed” by colonial rule, is much debated. See David N. Lorenzen, “Who Invented Hinduism?” Comparative Studies in Society and History, XLI (1999), 630–659. In the archival record, Parsi residents are usually denoted or self-identified by the overarching category of “Parsi.” However, a report of an intra-religious dispute drew a distinction between two sub-groups in the Parsi religion, referring to “the two tribes of Parsees, the Andiaroos or order of Priests, and the Parsees generally so-called.” Report of a Committee for the Examination of Disputes Amongst Parsees, 5/1/1786, Public department, 1786 no 88, part ii, msa. Similarly, for Muslims, sub-categories mentioned in this period include Arabs, Moguls, Sheiks, Syeds, Bohras, and so on. Use of the overarching term Mussulman, Mahomedan, or Moorman in petitions suggests that the term had social meaning.


For the relevance of names as religious data in South Asia, see Raphael Susewind, “What’s in a Name? Probabilistic Inference of Religious Community from South Asian Names,” Field Methods, XXVII (2015), 1–14. The Bombay Calendar and Register for the year 1821, printed in Bombay by Francisco de Jesus; Times of India Calendar and Bombay Directory for 1884, Times of India Press, Bombay; Samuel T. Sheppard, Bombay Place-Names and Street Names: An Excursion into the By-ways of the History of Bombay City (Bombay, 1917).


The names of neighbors were more difficult to classify than were the names of buyers and sellers; neighbor’s names were not recorded as carefully. Since multiple parties were never listed for a single neighboring property, borrowing information from other parties was not an option. Parsey or Parsee is the old spelling of Parsi. Bhori and Coja (modern spellings Bohra and Khoja) are Muslim sub-groups. Moreman, which implies Moorman, refers to a Muslim resident. A challenge to using names to extract religious affiliation is the presence of ambiguous cases—people practicing more than one religion, or religious conversion (including the time lag between converting to another religion and changing or maintaining a name after religious conversion), or names routinely shared across groups. The development of algorithms that assign probabilities to group affiliation is a topic for further exploration with regard to historical names, but it would not solve all the problems of name classification.


All property sales on the island were required by government regulation to be publicly announced at least three weeks in advance. See Committee of Buildings Meeting, 11/12/1790, Committee of Buildings Diary, 1/177, 1787–1793, msa. Bombay Courier, 15 (741), 11/29/1806160, ProQuest Historical Newspaper Archive, University of Oxford. “Late of Bombay Gentoo,” indicating a Hindu resident of Bombay who was deceased.


Women in colonial Jamaica also transacted property to raise cash rather than to invest. Reid and Ryden, “Sugar and Jamaica’s Property Sales,” 415.


The data set has no European buyers or sellers; sales of property to Europeans might have been registered elsewhere, though no separate European-only register has been located.


Chi square test is significant at 0.001.


Conversions could in theory create heterogeneity in overarching religious categories. Unfortunately, no information about the prevalence of conversion in early nineteenth-century Bombay, or about converts maintaining relationships with kin who did not share their religion, is available.


The benefit of the odds ratio is its independence from the overall composition of the different groups among buyers and sellers. From Table 1, it would be 27/12÷7/16=5. Twenty-seven is the number of sales from Hindu buyers to Hindu sellers; twelve from Hindu sellers to non-Hindu buyers; seven from the non-Hindu sellers to Hindu buyers; and sixteen from non-Hindu sellers selling to each other.


For a similar dynamic, see Desmond Fitz-Gibbon, “The London Auction Mart and the Marketability of Real Estate in England, 1808–1864,” Journal of British Studies, LV (2016), 295–319.


F-test statistic=3.62; p=0.062.


Sheppard, “Bombay Place-Names”; note from Dickinson, Revenue Surveyor, to Secretary to Government, 7/8/1812, Revenue Department Diary 79, 1812, msa; Coppersmiths’ case, Committee of Buildings Diary, 4/180, 1803–07, msa.


Chi square tests comparing homogeneity of neighborhoods in oarts with those for houses were significant across the three measures of homogeneity.


Logan and Bellman, “Before The Philadelphia Negro”; Masselos, “Social Segregation and Crowd Cohesion”; Petition, 10/31/1792, Committee of Buildings Diary, 1/177, 1787–1793, msa.

Author notes

The author thanks Carlo Barone, Khushnuma Daruwala, Julia Drew, John Logan, Saumya Roy, Jusmeet Sihra, Shabnum Tejani, and Michael White for generous help with methodology and early drafts.