10 | Enumerating ‘Fuzzy Communities’: Recording Caste in the 1881 Census

By Nishitha Mandava and Shivakumar Jolad

Published on: 14 May 2026

While it is well established that the political currents of a given historical moment shape the making of the census, it is equally important to recognise that the census itself can have bearings on the society, producing frictions and contestations over social identities. Communities were grouped, elevated, or diminished at a mere ‘statistical sleight of hand or administrative flat’ (Conlon, 1981). 

In this article, we turn to the 1881 census, paying attention to how caste and ethnic identities were classified and recorded. The stakes of this inquiry assume more significance when we recall that just a decade ago, in 1871, the colonial state passed the draconian Criminal Tribes Act, under which the communities the census named and counted could be branded as irredeemable hereditary criminals and subjected to surveillance, incarceration, and forced settlement.

Decades earlier, Company-era paintings like Seventy-two Specimens of Castes in India (1837) had already begun to catalogue the Indian subcontinent’s social world by moving away from painting imperial courts and princely grandeur to capturing everyday people. They fixed caste into visual types, each figure posed as a specimen, the underlying assumption being that a specific set of characteristics were representative of a specific caste. This assigned place for each caste in the ‘colonial sociological theatre’ (Cohn 1996). 

A similar ethnographic impulse of capturing details of Indian castes, their customs, professions and other mundaneities of their lifeworlds informed the census as it set out to record India’s social identities. 

Given the complexity and open-ended nature of caste and ethnic identities of India, scholars have often termed them as ‘fuzzy communities’ (Dirks 2001; Cohn 1987).

A dominant narrative that shapes our current understanding of caste enumeration holds that the census established definitive geographical and social boundaries for otherwise fluid identities, only to increase divisiveness. A proponent of this view, Bernard Cohn (1987) argues that while the census gave the British a panoptical view of Indian society, it also made language, caste, and religion as measurable and comparable units with defined boundaries. 

However, this view has not been without criticism. Sumit Guha (2003), among others, complicates this argument that the census conjured otherwise non-existent categories or gave highly amorphous identities a more defined form. A key takeaway from this debate is that enumerating social identities in India has and will always be a complex, arduous process, in which officials’ preconceived notions about India’s ethnic and social identities are often written into the census’s categories and data. 

At the same time, it is important to recognize that caste underwent a significant transformation under colonial rule, becoming a crucial site where Indian social realities intersected with colonial systems of knowledge and governance. As Edward Said (1978) rightly suggests, representations and classifications often intertwine with ‘great many other things besides the “truth”’ which is circulated, consumed, refurbished and in all this process it outlives the context which produced it, continuing to inform our contemporary understandings of the subcontinent’s social world. 

Caste has been used as a social marker in Indian society, and the categories often include Brahmins, Kshatriyas, Vaishyas, and Shudras, collectively known as the Varnas. Each Varna houses various castes and sub-castes, which differ from region to region, and the hierarchies are not always clearly drawn out and are liable to change over time. This was not simply a form of social stratification but a form of social inequality which attached certain notions of purity and impurity to each caste. Hence the term ‘caste’ itself originates from the Portuguese word casta, meaning ‘pure.’ 

B.R. Ambedkar, a canonical anti-caste thinker in his well-known critique, argued that the Caste system in India is not about division of labour, but  labourers. Caste permanently divided people into rigid, hereditary, and hierarchical occupational groups. He argued that such a system destroyed individual freedom and human dignity as it denied people the ability to develop their talents outside caste-prescribed roles.  Ambedkar (1917) observed that: ‘Endogamy is the only characteristic that is peculiar to caste.’ Meaning, while hierarchy and religion are characteristic to caste, it is the marriage restriction that sustains caste boundaries. 

The British census was not the first to record caste. An interesting example comes from the Arthashastra, an ancient Indian treatise on statecraft and political strategy estimated to have been compiled between 2nd BCE to 1st CE. Take another fascinating instance, the survey titled Marvar ra Parganam ri Vigat undertaken in 17th century  Marwar which provided a caste-wise breakdown of urban population. In this long genealogy of caste enumeration, the 1881 census understood caste in the following manner: 

‘The only classes of Hindoo religion are four, namely, Brahmin, Kshattriya, Vaisya, and Sudra. Dhers and Chamars are merely men of mixed class, and are not only an excommunicated set of people, but are held in great detestation and hatred by the three superior classes of people, who never allow the shades of the bodies of those outcasts to fall upon their food and drink or their bodies. The Code of Manu, and the Sudra Kamala and Kara, give a minute description of these castes, but they are evidently the Antyaja tribe, which term means “latest born.” The religious penance for killing them is the same as for killing a cat, a frog, a dog, a lizard, and various other animals.’ (Plowden, 1883). 

Since the 1872 census itself, the British incorporated categories of race, caste and religion, but the distinctions between these categories have remained murky. The caste category was applied only to Hindus until 1891, while the race and tribe categories were applied to religious groups that did not belong to the Hindu or Jain religions. The following table shows the evolution of how caste was recorded across each census:

Table 1 (Source: Bhagat, 2006; Srivastava, 1972) 

In this essay, using the 1881 census, we will examine how early censuses produced caste as a category of enumeration. 

The Scale of the Caste Census

The 1872 Census of British India was the first large-scale colonial attempt to systematically count and classify caste across the subcontinent. However, the census itself admitted that it failed to create a coherent system of caste classification because each province used different methods suited to local conditions. This resulted in extensive but fragmented findings. These administrative complexities and non-coherent nature of caste enumeration crept even into the census exercise undertaken in the following decade. 

The 1881 census aimed to provide detailed statistics, particularly for castes with populations over 100,000. Similar to the 1872 census, it relied on provincial officials to determine, based on their local knowledge, whether different caste names were synonymous and could be grouped under a common title, and to report any such combinations. 

It also sought to display the distribution of major castes by district, along with relevant details such as total numbers and the composition of each caste by sex. However, an all-India classification was not yet feasible due to incomplete provincial reports and the need for further expertise. 

An initial attempt to classify castes by social position was abandoned after it sparked petitions and disputes and was deemed too uncertain. Instead, the data was organised more simply into Brahmans, Rajputs, and ‘other castes,’ with the latter comprising a very large number of groups listed in detail. Additional castes with smaller populations, typically under 100,000, were included separately, often across multiple provinces. A total of 207 castes were identified, with an additional 65, bringing the total to 272 in India (Plowden, 1883). 

Table 2 Major Hindu Castes with Populations Exceeding One Million (1881) (Source: Plowden, 1883)

This table illustrates the primary castes identified by the 1881 Census that reached a significant numerical threshold across the Empire.

The total population covered by this census for caste enumeration amounted to 136,689,714, while a further 25,350,547 persons were recorded in provincial statements without caste specification and were therefore excluded from the all-India compilation. 

In addition, 3,124,553 persons are noted as belonging to castes that appear separately in more than one province, even where their numbers in any single province fall below 100,000 but exceed 50,000, and a further 2,118,750 persons belong to castes numbering less than 50,000 in any one province and appearing only in a single province (Plowden, 1883). 

When combined with the figures for Brahmans and Rajputs presented elsewhere, these data are understood to account for the majority of the Hindu population within the classificatory limits adopted. 

Out of a total Hindu population which was approximately 187.9 million (including Burma), Brahmins constituted around 7.3%, Rajputs roughly 3.8%, while the majority of it, nearly 89%, were classified under ‘other castes.’ These estimates should be treated with caution due to inconsistencies in provincial enumeration and classification practices. 

Table 3 Distribution of Hindu Population (Source: Plowden, 1883)

(Total Hindoo population, as shown in Table 3, is 187,937,450)

Another key limitation was that identical castes appeared under different names across provinces, for instance, Nai and Ambattan, or Ahir and Gopala, but such categories were not amalgamated due to inconsistencies in provincial reporting. The report suggests deferring standardisation until more complete and comparable data become available.

W.C. Plowden (1883), the then Census Commissioner, found this whole exercise so confusing that he recommended its total abolition from the census returns. In his Census Report of the North-Western Provinces, 1882, he observed: 

The colonial officials knew so little about castes that they thought it was wise to record dizzying amounts of detail. As a result, the classification of castes became, for Plowden, over-elaborate and hardly reliable.

Plowden, who had already despaired of caste enumeration had not grown more optimistic having overseen the collection of roughly 200 pages of caste descriptions and 1,370 pages of tables. He concluded that the information gained from it did not repay the trouble taken in the compilation.

Fuzzy Communities 

The 1881 Census attempted to classify the entire Indian population by religion, listing fourteen designated categories, including Hindu, Muslim, Christian, Buddhist, Jain, Sikh, Parsi, Jewish, Brahmo, and residual groups, which were placed under the broad category of the ‘aboriginal.’ 

Local authorities who acted as census takers in the earlier census were given ‘special keys’ for converting unsuitable responses into officially formulated census categories (Plowden 1873). This was to aid the Indian informants as they would often fail to place themselves in already prescribed columns of caste and religion, attesting to the blurred nature of the ‘categories’ or the relative insignificance of them in the everyday lives of people till then (Banerjee-Dube, 2014). 

The British officials interpreted this ‘failure’ on the part of Indians as them incapable of identifying themselves for administrative purposes. Despite such ‘special keys’ various difficulties arose. 

The first challenge were the sectarian groups which drew on multiple religious traditions. The Satnamis and Kabirpanthis, both born out of religious reform movements, had emerged from within the broader Hindu social world. The Kabirpanthis were followers of the 15th-century poet-saint Kabir, whose teachings drew on both Hindu and Islamic traditions and rejected caste hierarchy and image worship. At the same time, the Satnamis were a monotheistic order concentrated in the Chhattisgarh region who similarly rejected Brahminical authority and caste.

Despite their distinct theological orientations, both groups continued to observe caste, perform Hindu rites and ceremonies, and practice endogamy within their respective communities. The census report observes that Kabirpanthis were ‘Hindoos in almost every particular, and Kabirpanthis in nothing but name,’ and their entries were accordingly corrected to Hindu. This  correction essentially erased the very distinction these communities had historically drawn for themselves.

The second concern was that the local enumerator recording religion in the household was often himself a high-caste Hindu, and was not above his prejudices. Lower castes such as Dhers and Mangs were communities that were largely excluded from Hindu temples and Brahminical religious life. 

When upper-caste local enumerators were asked to record the religion of such groups, many refused to classify them as Hindu, instead noting the caste name itself in the religion column, producing non-existing designations such as ‘Dher religion’ or ‘Mang religion.’ 

In British Districts, orthodox Hindu enumerators were reluctant to record hill tribes as Hindu, whereas in the Feudatory States, where aboriginal communities were numerically dominant, their religious identity was more readily recognized. The same population, in other words, could be classified differently depending on where they lived and who was doing the counting. 

Plowden’s report itself acknowledges that these people were probably not even asked about their own beliefs. Some enumerators drew a distinction between high-caste Hindus who worshipped carved images and low-caste communities who worshipped ‘daubed stones.’ At the central office editorial interventions were made and they were reclassified as Hindu unless recorded under another recognised religion. 

The third difficulty was ascertaining which were main and which were sub-castes. In the case of Madras, the Caste Index listed groups under the headings of ‘major’ and ‘minor.’ For instance, entries would be ‘Brahman, Saraswat’ or ‘Brahmin, Other’ but many were indexed as ‘Other, Other.’ Among Hindus, a striking 47% were not classified within their minor divisions (Conlon, 1981). It is evident that caste identities often exceeded the classificatory categories of the census.

The fourth difficulty was the ambiguous ‘aboriginal’ category, which referred to tribal populations that had not converted to Christianity, Islam, or Hinduism and practised what the report described as the ‘primitive cult of their forefathers, adoring nature,’ encompassing the worship of nature and local deities. The Deputy Commissioner of Ellichpur (Maharashtra) recounts questioning Korkus, Gaolans, and Nihals about their religion. 

When asked, the Korkus to name the gods they worshiped, one of them replied: ‘I am a Korku, but I do not know what my religion is called. I worship Mahadeo, Hunaman, Byram-Bai, Chand, Suraj, and the Bhagwant, who is the author of my religion, call it what you please.’ When pressed further, five of the six Korkus answered ‘Hindoo,’ apparently as the closest available name for something they themselves could not categorise (Plowden, 1883). 

Colonial eyes classified religious practices that did not map onto recognisable world religions as primitive and prior to them. Interestingly, the Bengal authorities found this category too broad and undertook a more detailed examination of the tribal populations in their province. 

Provincial Reports on Caste 

The census was a sprawling administrative exercise that produced several provincial reports, written by different superintendents with varying methodological approaches. For the 1881 census, this included the Imperial Census of 1881 digest for Bombay and Sind, J.A. Bourdillon’s Report on the Census of Bengal, and Denzil Ibbetson’s Outlines of Punjab Ethnography. 

The Bombay digest engaged most directly with caste as an enumerative category. With 84% of the Bombay Presidency recorded as Hindu, caste was an unavoidable organising principle for the data. The digest produced a table showing class and name of caste alongside hereditary occupation (‘Caste in India in 1881,’ 1884). The largest caste recorded was the Kunbi, or cultivators, of the Maratha districts, followed by the caste Mahar and Dhed, described as village servants. Brahmins and Rajputs were noted as leading socially. Both Mahar (to which Dr. B R Ambedkar belonged to) and Dhed were the traditional untouchable communities.

It is also worth noting here how such classificatory exercises continue to have significance to the present. In 2025, the Maharashtra government accepted the 1918 Hyderabad Gazette as the basis for issuing Kunbi caste certificates to Marathas in the Marathwada region, enabling access to Other Backward Classes (OBC) reservations. 

Bourdillon’s Bengal report took a more conservative approach. He largely avoided elaborating on caste beyond what was strictly necessary for tabulation, and the Bengal tables dealt only with ‘Hindu castes’ as a category. 

Denzil Ibbetson (1883), the Census Commissioner of Punjab contested the popular and ‘currently received’ theory of caste as a purely Hindu religious institution, arguing instead that it had no necessary connection with religion and that conversion from Hinduism to Islam had not resulted in the slightest effect upon caste. 

For Ibbetson, caste in Punjab was largely based on occupation, and among the mass of land-owning and cultivating people, upon political position and tribal identity. He was inspired by another colonial official, John Nesfield’s materialist evolutionary idea of castes which understood them as social guilds that descended from tribes, where ‘function’ provided the main criterion of classification (Banerjee-Dube, 2014). 

Ibbetson proposed a five-stage theory of caste’s evolution: from primitive tribal division, through hereditary occupational guilds, to the exaltation of priestly and Levitical authority, culminating in the Brahminical rules and rituals governing marriage, food, and social intercourse.

Insights from these three reports demonstrate that there was no consensus among colonial administrators about what caste actually entailed, whether it was religious, racial, occupational, or tribal in origin. Unsurprisingly, they struggled to come up with a consistent basis on which it could be recorded. 

Marriage and Civil Conditions

Caste was also employed to interpret marital data. The census of 1881, lists 17 distinct Hindu castes, along with their percentages of single, married, and widowed persons. The report observes that variations between castes in marital condition were smaller than ‘popularly supposed,’ and that Christians and Muslims showed figures with a distinctly ‘Hindoo tinge’ (Plowden, 1883). Two-thirds of Muslims and most Christians were Hindu by descent, tradition, and manners. 

Data on Brahmins were used to estimate the average age at which a Brahmin woman married, revealing that nearly all Brahmin girls married before age 10. The data on Pariahs, interestingly, demonstrates the smallest proportion of widows and the largest number of unmarried women. The census attributes this to lower rates of infant marriage among lower castes.

The Limits of Education

The 1881 Census report explicitly states that educational statistics by caste were not dealt with in the All India Tables. While data was collected by religion for the entire Empire, the correlation between caste and education was only explored in certain provincial reports, such as those for Bombay and Berar. The following table, compiled by the official, E.J. Mr. Kitts for the Berar Province, included in Plowden’s report shows the number of boys from the ‘more important’ Hindu and Jain castes attending government schools at the end of February 1881:

The Kunbi, as one of the largest cultivating castes, sent the highest number of boys to government schools, 6,129, but this represented only 6% of their school-age boys, reflecting the size of their population rather than an educational advantage. 

The text accompanying the table further clarifies the access different castes had to education, painting a rather bleak picture. Of the 22,176 Hindu and Jain schoolboys attending government schools in Berar, 95% belonged to the castes listed in the table, leaving the remaining hundreds of Hindu and Jain castes accounting for only 5% of school attendance. 

The Spectre of the Thuggee 

The Thuggee was said to haunt the highways of India, preying on unsuspecting travellers, robbing them and then strangling them to death, earning their practitioners the name phansigars, the noose-men. It was a crime that seized the British imagination with particular force. Against the backdrop of Orientalist fantasies of the British officials, these criminals were portrayed as a pan-Indian network belonging to hereditary religious cults that worshipped Goddess Kaali (Cole, 2001). 

In a broader sense, the colonial state interpreted mobile populations such as dacoits, fakirs, goondas and herders as threats to the prescribed sociological order as they were often on the move and were not held down by settled agriculture or waged labour (Cohn, 1987). These groups due to their unpredictable mobility were interpreted as beyond the civil bounds. 

They were subjected to investigations, and large groups, tribes, were stigmatised as criminal tribes. Such activities were carried out with renewed aggression with the establishment of the Thuggee and Dacoit Department in 1835. A number of people were apprehended as thugs, some guilty, some innocent (Cole, 2001). 

In 1871, the Criminal Tribes Act was passed, under which hundreds of castes could be classified as criminal. Meaning by the virtue of being born into a particular caste one could be presumed to be criminal. This resulted in the police gaining unchecked power to control their movements and the individuals of these tribes were expected to obtain a pass for their travel and periodically report their movements to officials (Cole, 2001). 

The census was not immune to these notions of criminality and social communities in India; the spectre of the Thuggee lingered within it. In describing a ‘criminal tribe’ in Berar, the report stays true to the orientalist tropes of perceiving this individuals as part of religious cults that worshipped goddess Kaali and belonged to ostensibly hereditary criminal castes: ‘The Ran Kaikaris, or Kul Korwas, are the most criminal class; they act under a chief who is elected for life […] The Kaikaris worship Bhawani and often carry with them a small image of the goddess which they invoke in fortune telling.’ 

In an earlier instance the report notes them as ‘notorious in the Bhandara district of the Central Provinces as determined and skilful thieves’ (Plowden, 1883). 

The 1881 census for castes in Punjab utilised categories like ‘gipsy’ tribes to make note of communities that were mobile and ‘wandering,’ drawn from Ibbetson’s report on Punjab: 

Table 5 Communities in Punjab (Source: Plowden, 1883)

It also provides a specific list of ‘wandering’ and ‘criminal’ tribes in Punjab:

Table 6 List of ‘wandering’ and ‘criminal’ castes in Punjab (Source: Plowden, 1883)

It is essential for us to recognise that terms like hereditary, criminal, and wandering were not ahistorical labels but were part of a specific history of colonial paranoia and criminal anthropology which the census gave numerical expression to. Of course, criminal tribes were not entirely colonial constructs, we find references to communities of bandits across Buddhist, Brahmanical, Mughal and Early European sources (Piliavsky, 2015). The category of criminal tribe is a product of complex genealogies, wherein the colonial rule institutionalised  and surveilled these communities in unprecedented ways. 

Upon independence, acts like the Criminal Tribes Act were repealed in 1952 and the criminal tribes were re-classified as Denotified Tribes. But colonial stigmas against them continue to shape how they are perceived in contemporary India, particularly affecting their access to education and employment (Devy, 2021). 

Conclusion 

When the 1881 census set out to classify and count castes in India, the colonial officials did not fully realise the messy and taxing endeavour that it would turn out to be. Across provinces, enumerators disagreed on what caste was, what it belonged to, and how it should be recorded, demonstrating that a uniform colonial discourse did not exist. The same community could appear under different names, different religions, or different moral designations depending on who was counting and where. 

While this did not produce a faithful portrait of the subcontinent’s social world, it provided crucial insights into how caste and other social identities interacted with each other to structure an individual’s life including facets like marriage and education across the subcontinent. One could argue that it offers a truth but not the truth. 

This had economic and political implications especially with some castes being labelled as ‘martial races’ which made them readily sought after for colonial army recruitment while some were labelled as ‘criminal tribes’, subjecting them to constant colonial surveillance and suspicions. 

Communities increasingly recognised that enumeration could shape access to representation, employment, and state patronage. By the late nineteenth and early twentieth centuries, associations and reform movements were petitioning census authorities to alter or elevate their official status. 

As we will explore in the articles to come, caste as a census category has had a long history of providing striking insights into the marginalization of the lower castes. Consider the first census of Madras, which found that Brahmins dominated the learned professions, while personal services were largely performed by lower castes (Cornish, 1874) or the 1918 Hyderabad Gazette as a basis for Kunbi claims to OBC reservations. Such insights in the 1920s and 30s formed the basis for anti-caste agitations and for lower-caste claims for increased political representation when reforms for self-governance were being introduced through the 1919 Montagu-Chelmsford Reforms and the series of Round Table Conferences in the 1930s (Singh, 2021). 

Such histories transformed what had begun as a confused exercise into an instrument of the colonial state through which it came to know, surveil and govern its subjects, while these same subjects utilized it to articulate their claims to rights, status, and political representation. The afterlives of colonial enumeration continue to haunt us even today. Contemporary debates surrounding caste censuses, reservations, and demands for greater statistical visibility among marginalized groups continue to draw upon categories first formalized under colonial rule. 

Today, as we imagine, classify, and make sense of social identities in India it is worth reminding ourselves how it continues to bear the imprint of exercises such as the 1881 caste census. The categories forged through colonial enumeration long outlived the moment of enumeration itself, shaping governance and representation, but also our everyday language and assumptions through which we have come to understand communities. 

(Authors: Nishitha Mandava is an independent researcher and research Consultant for

Center for Legislative Education and Research, FLAME University, Pune;

Dr. Shivakumar Jolad is the Chair of Center for Legislative Education and Research and Director of India State Stories, FLAME University, Pune

Nishitha contributed to research and primary writing;  Shivakumar contributed to conceptualization, research, visualizations and editing)

References

Ambedkar, B. R. (1917). Castes in India: Their mechanism, genesis and development. Indian Antiquary, 41, 81–85. (Original paper presented 1916).

Banerjee-Dube, I. (2014). A History of Modern India: (1st edn). Cambridge University Press.

Bhagat, R. B. (2006). Census and caste enumeration: British legacy and contemporary practice in India. Genus, 62(2), 119–134.

Caste in India in 1881. (1884). [Review of Outlines of Punjáb Ethnography; Imperial Census of 1881; Report of the Census of Bengal, 1881, by Denzil Charles Jelf Ibbetson & J. A. Bourdillon]. Science, 4(98), 557–558. 

Conlon, F. (1981). The Census of India as a Source for the Historical Study of Religion and Caste. In The Census in British India New Perspectives (pp. 103–118). Manohar Publications.

Cohn, B. S. (1987). The census, social structure and objectification in South Asia. In B. S. Cohn, An anthropologist among the historians and other essays (pp. 224–254). Oxford University Press. 

Cohn, B. S. (1996). Colonialism and its forms of knowledge: The British in India. Princeton University Press. 

Cole, S. A. (2001). Suspect identities: A history of fingerprinting and criminal identification. Harvard University Press.

Cornish, W. R. (1874). Census of the Madras Presidency 1871. Government Gazette Press. 

Xaxa, A., & Devy, G. N. (Eds.). (2021). Being Adivasi: Existence, entitlements, exclusion. Penguin Random House India. 

Dirks, N. B. (2001). Castes of mind: Colonialism and the making of modern India. Princeton University Press. 

Guha, S. (2003). The Politics of Identity and Enumeration in India c. 1600-1990. Comparative Studies in Society and History, 45(1), 148–167. 

Ibbetson, D. (1916). Panjab castes. The Superintendent Gov. Printing, Lahore.

Plowden, W.C (1883). Report on the Census of British India taken on the 17^th February 1881, London, Eyre and Spottiswoode.

Plowden, W. C. (1873). Census of the North-Western Provinces 1872: General report (Vol. 1). Allahabad. 

Piliavsky A. (2015). The ‘Criminal Tribe’ in India before the British. Comparative Studies in Society and History, 57(2), 323-354. 

Said, E. W. (1979). Orientalism. Vintage Books.

Singh, P. (2022). Numbers as a Means to Power: Politics of Caste as a Census Category in Colonial India c 1871–1941. Economic and Political Weekly, 57 (2022).

Srivastava, S. C. (1972). Indian census in perspective (Census Centenary Monograph No.1). Office of the Registrar General, India.