Politics of Population Census Data in India Though the census count of human population seems to be an apolitical affair, yet it is invariably coloured by political overtones of various hues. Accordingly, both the detail and the quality of population data carry an overt as well as covert imprint of political considerations. Thus, for a proper comprehension of population patterns, both social and spatial, it is imperative to understand the politics of collection of such data. This paper attempts to understand mainly the following aspects of politics of population census data in India: (i) categorisation of census materials; (ii) changes in census categories over time; and (iii) changes in details of data by caste and religion. MEHAR SINGH GILL

A

census of the population is not only the most extensive and most public activity related to population data, it also provides the largest amount of data used by researchers, policy-makers and administrators [Bulmer 1979: 158]. In the case of India also, the census “is far the most important source of demographic data” which is widely used for social and economic planning in the country [Bose et al 1977:19]. Census enumeration in each country is done mainly with two objectives in mind: (i) to record facts about demographic and other related socio-economic attributes of a population, and (ii) to construct a characteristic demographic and socio-economic space by introducing new categories and/or replacing the earlier ones for the purpose of data collection. Each social and political system employs various categories to designate socio-economic and political contours as it wishes to articulate and promote them. The nature and detail of data on ethnic, religious and other minority groups is chiefly determined by the current political discourse of a country. If the discourse aims at accommodating differences along various ethnic religious and tribal lines, then the population data would also reflect such differentiation. For instance, the data issued by the latest census of Ethiopia are available for 80 ethnic groups excluding foreigners [Census of Ethiopia 1998]. The census furnished data even for communities with population less than 100 each at the country level, e g, Mabaan who numbered merely 23, and Gebato who had a population of 75 only. Similarly, at the regional level tribal group(s) with a population of one person only were given separate representation in the Census of Ethiopia. On the other hand, sometimes emergent political consideration may require expressing a largely homogeneous picture of population as is being done in India and many other countries of the world. If a community is omitted in census counts, its socioeconomic and political visibility, as a distinct entity would gradually get lost to the socio-political discourse of the country. In quite a few cases, even the veracity of data would be of suspect nature. This is particularly the case in areas marked by ethnic tensions, latent or otherwise, and prolonged armed ethnic conflicts. Similarly, the terminology employed by the census organisations has its own politics and meaning. For instance, the people who were classed as nationalities in the former USSR, are now being referred to as ethnicities in the post-Soviet Russia [State Committee of the Russian Federation on Statistics 1998].

Economic and Political Weekly

January 20, 2007

Foucault (1980, 52) rightly emphasised that power and knowledge are closely interconnected. Categorisation of people and the information generated about such categories in a country play an important interconnecting and supporting role between power and knowledge. These categories provide a characteristic legibility to a society by strongly influencing perceptions of us and the Others. No wonder that each political power generates information/data and knowledge in terms of its own interests and worldview. This is the reason that no country has evinced interest so far in collecting and publishing data on human rights abuses in its own territory as it would reflect adversely on its image. The obtaining complexity of social reality necessitates categorisation for its better comprehension. In fact, “the human social world might not be manageable at all” in the absence of categorisation [Jenkins 2000:8]. In common perception, population data collection is typically an apolitical affair covering numbers, demographic traits and other socio-economic characteristics of the people. However, when viewed closely it becomes clear that, to a notable extent, the collection of census data is determined by political considerations. Which categories are to be made or unmade is invariably strongly influenced by political factors. People who wield power as well as those who are capable of resisting the effects of power play the prime role in deciding social and spatial categories employed for the census and for administrative purposes. In quite a few cases, changes in census categories over time also manifest changing definitions of us and the Others and the nature of relationship between the two. In a way, census categories not only reflect social identities, but also help in creating new identities in tune with specific discourses of major political parties in a country [Sundar 2000:117; Nakarda 1983:379; Bhagat 2001:4352; Bhagat 2003:689; Cohn 1987:250; Smith 2000:5-6]. Such categories not only define different social strata, but also play a part in “locking in” individuals to specific groups [Christopher 2002:406]. Expectedly, the census data by different social categories is employed for socio-economic assessment of and planning for different sections of people. In this context, these census-created social classes become important for various levels and shades of politics. Thus, apart from other requirements, census data in each country are also generated in order to create/enhance/erode a specific type of legibility of the society. For instance, in place of a number of “untouchable” castes

241

a new category called “scheduled castes” has been created through the Constitution of India. The creation of scheduled caste category is highly ideological and political, which strengthens the majoritarian view in the country [Escobar 1997:186-187]. If it suits the ruling class(es) of a country to subdivide the Other, then a large number of census categories could be used as was done under the British rule in India. On the other hand, if the purpose is to assimilate the Other, then even its key distinguishing features or differences are either ignored or denied altogether as has been done particularly in case of the Jains, the Sikhs, the Buddhists and tribal people in the country since 1947. It is well known that in each state, the dominant people weave various sorts of cocoons around those who are considered Others so that they get hidden from the view [Spicer 1994:39]. The use of characteristic census categories is one important method of cocooning such people. Thus, making and unmaking of census categories lead to doing and undoing of social and spatial boundaries. Not only are data made available through these divisions, but these also provide social and spatial contexts for analysing population phenomena. Besides, such socio-spatial boundaries further help in (re)creating various communities demographically, sociologically, and culturally [Escobar 1984:384]. Further, competition for jobs and other perceived privileges along these lines play its own role in strengthening census-created categories, which in turn, become an essential part of politics [Nakarda 1983:379]. In other words, census categories provide important coordinates for production of information and knowledge, which are necessary for the construction and maintenance of states.

Census Data in India With a long history of more than 130 years now, the Indian Census is one of the most comprehensive in the world. As expected, the census has also undergone notable changes in categorisation as well as details of data. Under the British rule till 1947, the Indian Census Organisation had employed many more categories for census purposes as compared with those in the post-British period. The differential in the number of categories used in this regard is a manifestation of the basic difference in the political discourses of the ruling classes during the British and the post-1947 periods. The political agenda since the departure of the British in 1947 has been essentially homogenising which also gets amply reflected in notable reduction in the number of ethnic/religious groups as well as the number of population attributes of religious communities for census taking during this period. On the other hand, the British rulers of India had no political compulsion, like that of the Indian state now, to go for a homogenising discourse. In fact, it suited the political interests of the British rulers to highlight more and more seams in the Indian society. Hence, for their administrative reasons as well as in antiquarian interests [Barrier 1981, viii], they opted for collection of population data for many more groups of people as separate entities, whether these were religious, tribal, caste or other groups. In order to increase/decrease socio-economic visibility of a community through census counts, the ruling class usually adopts three methods. One, selecting enumeration categories in such a way that only those attributes of people/population find mention, which suit the reigning political class. Two, providing more details of data and more space for discussion on various socio-economic

242

and demographic attributes of the “mainstream” political group(s). Several examples to this effect could be cited from the successive censuses of India both during the British and postBritish periods. For instance, in the volume entitled Report of the 1911 Census [Gait 1913], the discussion on the Hindus covered 11 pages, on the Muslims four pages, on the Jains one page, and on the Sikhs a half page only. On the other hand, the Christians with a population of 3,876,203 only in 1911 were allocated nine and a half page notwithstanding the fact that their number was merely 1.79 per cent of the Hindus and 5.88 per cent of the Muslims in the country at that time. During the British rule, the Christians were given much larger space in the censuses accounts as they were, in a way, a part of the “us” for the British rulers. Similarly, the census question on race during that period could also be partly attributed to the then ruling elite’s strong interest in us vs the Other dichotomy. Three, carving out census and administrative areas in such a manner that concentrations of certain people get variously bifurcated. In India, all the three methods could be found at work regarding the collection of population data. For the censuses conducted during the British rule, detailed data were collected about different religions, castes and tribes in the country. For instance, the 1931 Census [Hutton 1933] had recorded literacy rate for 138 castes. It had also published data on literacy by religion and age for several religious communities and sub-communities: Hindus, Muslims, Sikhs, Jains, Christians, Zoroastrians, Buddhists, Jews and tribals, etc. Similarly, a large number of tribes had been focused upon for census purposes in the British India. According to one estimate [Singh 1996:140], census data were collected pertaining to about 3,000 communities in India during the British rule. In the process, many tribes acquired social and academic visibility which otherwise would not have been possible, at least to the extent it came. The British rulers of India had three main objectives in detailed census taking: (i) to understand the demographic and social structure of the country which was essential for strengthening the colonial grip over the people [Singh 1996:138]; (ii) to acquire better comprehension of the resource potential, both demographic and cultural; and (iii) to enhance the demographic visibility of the Christians in the country as this community was given a separate column in all the censuses in various areas of the country irrespective of its relatively small numerical strength in India, particularly in the 19th century. For example, though the Christians had a nominal population of 4,864 only in the urban centres of the present-day Punjab area in 1891, yet they were treated as a separate population group for census purposes in that year also. Many people in the country hold the view that the British rulers published these detailed data about various religious, tribal and caste groups for the purpose of creating divisions among the Indian people [Das 1994:121; Bhagat 2001:4353-55; Bhagat 2003:687; Bose 2005:371]. According to this view, the colonial rulers wanted to highlight various seams along which different population groups of the country could be easily split apart or pitted against one another. Though it might have been one of their objectives, it will not be wholly correct to say that collection of census data on various religions under the British rule was done with the sole objective of promoting religious communalism in India. In fact, the data on castes and religions have not only worked to enhance the cohesiveness of each one of these communities across the country, but, as a consequence, have also strengthened the territorial unity of the country. However, it

Economic and Political Weekly

January 20, 2007

cannot be denied that census taking during the British period did contribute to crystallise identities of quite a few “fuzzy”: communities at that time [Kaviraj 1993:26]. Similarly, it is also equally true that the post-indepedence censuses have worked to create fuzziness about many ethnic communities by denying a separate category for them in its schedules. If one accepts the view that the only motive of the British rulers, as has been commonly alleged, in publishing detailed data on religion was to create a division among the Indian people, then the same view should hold good regarding the BJP-led government of the country for allowing the tabulation and publication of notably more detailed data on religion from the 2001 Census, and also the Congree Party-led UPA government under wholse tenure it was released in September 2004. Significantly, it is for the first time in the census history of the country that “data on regilious communities has been cross-classified by literacy, sex ratio, child sex ratio (0-6 age group) and workers (for major categories)” by individual districts, states/union territories and rural/urban residence [Bose 2005:371]. However, it is notable that no research article has pointed out so far that the detailed data on religion from the 2001 Census has been published to create fissures in the Indian society. This differential response by academicians to the same type of data recorded in the pre- and post-independence period underline Facauldian view (1980, 52) that production of knowledge and circuits of power are closely intertwined. It is well known that enumeration of detailed demographic and socio-economic data on various religious and ethnic groups is a prerequisite for their in-depth understanding and proper socioeconomic planning. In this context, notwithstanding its varied interpretations and resultant implications, the Census Organisation of India deserves credit for publishing detailed data on religion from the 2001 Census after a lapse of six decades. The same credit is equally due to those who published similar data in the pre-independence period. Each state indulges, directly or indirectly, in erasure of difference and diversity within its territory, which is well in tune with totalising character of power. In order to achieve this objective, the ruling group employs various methods, which lead to “the normalisation and standardisation of reality” [Escobar 1997:178]. India is no exception in this regard. Since the end of the British rule in 1947, the basic anxiety of India has been to integrate and homogenise highly diverse parts of the country. Significantly, the ruling class of the country has adopted brahminical Hinduism in various shades, including caste system, camouflaged as “mainstream” culture as the master thread in this regard [Pandey 1998:3; Inden 1990:86]. The concept of Indian “mainstream” is a part and parcel of the process of standardisation of the country’s socio-culturally diverse reality. The census categories have made their own contribution in this direction. Almost all the major national political parties, particularly the Congress and the BJP, along with main parties of the Hindi belt, aim at erasure of socio-cultural difference in the country. Their political discourses privilege all-India structures over regional, and the mainstream over the “marginal” [Pandey 1998:29]. The main difference between the Congress and the BJP, in this regard, relates to the extent of socio-cultural homogenisation to be achieved and the strategy to be followed to achieve this goal within the overarching structure of Hinduism. The BJP, at least in the present context, has a design to “efface all the signs of non-Hinduness” [Khilnani 1997:189] in an aggressive and

Economic and Political Weekly

January 20, 2007

hegemonic manner. On the other hand, the Congress Party’s discourse aims at subordination of the Other to tame down any effective opposition to the mainstream thesis. However, the Congress has been mostly doing it in a subtle fashion. Significantly, whenever and wherever it suited to enlarge its vote bank, the Congress Party has also been equally open to adopting aggressive Hindutva agenda for a limited period of time [Brass 1992:16; Jaffrelot 1996:330; Khilnani 1997:183]. The widespread anti-Sikh violence for three days in Delhi, and most of other Congress-ruled states in the aftermath of assassination of Indira Gandhi in 1984 could be cited as a typical example in this connection. During the last about two decades, the Congress Party has also played a role that aids the rise of majoritarian nationalism [Ananth 2001:49]. Thus, both these dominant political parties as well as other major political parties of the country seem to be converging in the basic thrusts of their respective political discourses. Now even the Left parties have gone much distance with the mainstream thesis, which is essentially a majoritarian viewpoint, practised by these two dominant political parties [Banerjee 2003:864]. No wonder that with the fast growing emphasis on a seamless nationalism “the notion of inclusion as given in the Constitution is being substituted by the idea of exclusion” [Bava 2003:277]. It could be safely said that the ruling class of the country is so diffuse and subtle that it can get its agenda of socio-cultural homogenisation accomplished through whichever the political party or group of parties forms the central government. Population data visibility is also affected by characteristic delineation of census/administrative areas used as units of data aggregation. The boundaries of most of such areas have been strongly gerrymandered to serve the political agenda of the state. For example, most of the areas of tribal concentration, having more than 50 per cent share of tribal to total population, in central India have been variously bifurcated to fall in three or even four states as per the 1956 linguistic reorganisation of the states in the country. The main purpose of this type of gerrymandering was to dilute demographic and, hence, political strength of tribals so that they would not make up a sizeable share of population capable of making any meaningful political impact in any of the concerned states of the country. However, since early 1990s both the Congress Party and BJP have come to realise that instead of opposing regional sentiments directly, these could be harnessed to political benefit by creating smaller states. The emergence of three new states of Chhattisgarh, Jharkhand and Uttaranchal was due to the changed political perspectives of the country’s major political parties in this regard. The politics of the country has become so competitive that even “a few seats either way can now topple state and even central government” [Mawdsley 2002: 132]. Accordingly, now even the smaller regional parties find themselves capable of tilting the balance of power for or against the central government. In this context, the change in attitude of the major political parties could be called a part of invention and confirmation of truth, which is a major mechanism for the permeation and “the legitimation of hegemonic forms of power” [Escobar 1984, 392]. Moreover, India is also finding itself more confident to fight back any separatist movement as it has quite successfully done in many areas in the past few decades. Besides, as the “tribal” areas have already got flooded with non-tribals during the last four-five decades, there remains little danger of emergence of politically assertive tribal discourse(s) at present. In this context, it is not

243

unlikely that some more new states might emerge both in the tribal and non-tribal areas of the country. As in many other third world countries, sensitive census categories are made/unmade by the Indian government without any significantly meaningful academic or public debate [Christopher 2002: 402]. Invariably, any decision in this regard is often a political decision made by the top leadership of the country. Any seminar/conference, which might be held in this regard, is often to provide an academic legitimacy to a prior political decision. One instance would suffice to clarify the issue. In his opening address at the Census Conference held from February 23 to March 1, 1950, the then deputy-prime minister of India Sardar Vallabhbhai Patel announced that caste “will no longer be a prominent feature in the 1951 Census” [Patel 1950: 266]. It is also revealed by the fact that no political party in the country has cared to comment upon the census categories so far. In such a background, occasional dissenting voices by individuals, howsoever logical, do not matter much for census purposes in the third world countries like India. Some of the census categories as adopted in the post-British India fall in line with the state’s essentially linear narrative. Sensitive census categories not only reflect such a narrative, but also work to strengthen its homogenising thrust still further. For instance, census data by scheduled caste category has helped the state to subdue to some extent the forces of regionalism by opening up a new option for a section of people to organise themselves by this caste group which cuts across virtually the whole length and breadth of the country. With the change in the political discourse in the post-1947 period, there has also occurred a notable change in the nature of population data collected in the country. The census enumerations have “shown complete indifference to key cultural ethnic variables like mother tongue and religion in the planning of general economic tables, social and cultural tables relating to school/college attendance, and migration tables” [Mohanty 1996: 165]. With the sole exception of the latest population Census in 2001, data on religion have been limited to the count of males and females only, while those on castes have been notably curtailed. Significantly, beginning in 1951 for five successive censuses no data were published on occupational/industrial structure, and literacy rates of various religious groups, which are so essential for a meaningful understanding of demographic structure of any area/people. Moreover, there has been an increasing delay in the publication of data on religion, i e, 1971 Census data on religion were published just after one year in 1972, the 1981 data after three years in 1984, and those from 1991 Census were published after five years in 1996. The growing lag between the census enumeration and the publication of data on religion, simultaneous with increasing computerisation by the Census Organisation, could solely be attributed to political considerations. Bose (1991, 21) rightly says that “it is only a matter of time before such monitoring leads to manipulation of data”. The same could be said about data on language in the country. Similarly, data on individual scheduled tribes, save a few exceptions, are not made available. It is certainly an encouraging sign that religion data from the 2001 Census has become available in about three years time from the date of census enumeration. Though it is often emphasised that the focus of the postindependence census in the country is on economic and demographic parameters, yet even these parameters remained ignored in all the censuses from 1951 to 1991, regarding data on religious

244

communities in the country. The same deficiency could be observed regarding data about individual tribes since 1951. It is not easy to accept the view that such details of data are difficult to provide, as the census directors of different states remain so busy with other data that they “find it difficult to attach due importance to ethnographic data” [Srivastava 1996: 210]. In fact, availability of quite detailed data on religion from the 2001 Census has squarely falsified this view. It is generally suggested that data on literacy, education and occupational structure for different religious and tribal groups are not published as these can be used by vested interests to create/ enhance tensions along these lines. Now the question arises whether the tensions along caste lines are less important than those along religious and tribal lines? Again, the answer lies in the spirit of the narratives of the major political parties in the country. As caste system is an essential ingredient of Hinduism [Davis 1951: 162], any information, data or even interpretation along caste lines plays its own role to enhance visibility of the Hindus who constitute a very heavy majority in the country’s population. Although caste data go against the idea of monolithic Hinduism, yet distinguishes it from the other religious minorities which, expectedly, are and would continue to be viewed as the Other in the obtaining socio-political context in the country [Das 1992: 78]. However, this self-versus-other dichotomy is not something unsual as, historically, every society has been having its Other [Said 1994: 10]. Sometimes, the attitude of the respondents might also get biased by political and other social factors, which in turn, could adversely affect the quality of population data due to, what Kulkarni [1991: 207] calls, “fraudulent responses”. For instance, at the time of 1951 Census of India, owing to political and communal reasons, a large majority of the Hindu population of Punjab, along with a section of the rural Sikh scheduled castes, had been strongly mobilised to incorrectly record Hindi instead of Punjabi as their mother tongue. Consequent upon high Hindu-Sikh tension on this issue at the time, the 1951 Census did not publish language data for the then Punjab, PEPSU, and Himachal Pradesh. Similarly, language data in Punjab from the 1961 Census were highly vitiated as the population of the state stood divided along religious lines on this issue [Schwartzberg 1981: 51; Schwartzberg 1978: 235; Singh 1999: 225; Brass 1992: 163]. This happened in the wake of the agitation, launched by the Akali Dal, a predominantly Sikh party, in the 1950s for the reorganisation of the state on linguistic basis. Such reorganisation was strongly opposed by two “mainstream” parties, the Congress Party and the then Bhartiya Jan Sangh, later on rechristened as Bharatiya Janata Party (BJP), to which a very large share of the Hindus in Punjab belonged at that time. Accordingly, most of the Hindus in the state returned Hindi as their mother tongue notwithstanding the fact that they spoke Punjabi in their homes. Owing to the momentum generated at that time against Punjabi, even language data from the 1971 Census could not be taken as fully correct. For instance, the proportion of Punjabi and Hindi speakers in Punjab was 79.49 per cent and 20.01 per cent, respectively in 1971, whereas the corresponding figures in 1991 were 91.14 per cent and 7.20 per cent. From the data one gets the impression that the share of Punjabi speakers has gone up considerably at the expense of Hindi speakers during 1971-91. Although there has been a clear trend in the reverse direction, the notable increase in Punjabi speakers in the state was due to the fact that in 1991 the Punjabi Hindus returned their mother tongue more correctly which was not the

Economic and Political Weekly

January 20, 2007

case at the 1971 Census, and much less so in 1961. Similar instances of biased responses due to political considerations could also be found in the British Indian censuses [Schwartzberg 1981: 49-50].

Census and Religion Religion has been a regular and an important item in Indian census since the first census in 1871 [Jones 1981: 78]. However, the details of religion data have varied over time primarily in tune with the socio-political agenda of the rulers. Apart from their effect on the details of data on religion, political factors also played a significant role regarding accuracy of census returns in this regard. For instance, at the 1941 Census, there were complaints about biased enumeration of religious communities in Punjab and Bengal. The Hindus alleged that the Muslim enumerators had underenumerated the Hindus, while the Muslims had the same complaint against the Hindu enumerators for undercounting the Muslims [Maheshwari 1996: 126]. The religion of tribal people has also been a difficult issue for census purposes. For instance, the British Indian Census category “variously labelled as “primitive”, “animist” and “tribal”, which counted around 2.5 to 3 per cent of the population has been counted as Hindu in the post-British period [Oommen 2001: 228]. Now the question arises whether all these people have got converted to Hinduism since 1931? The answer to this question would be certainly in the negative notwithstanding the fact that there has also been a gradual trickling in of tribal people to Hinduism since 1947. Consequently, accuracy of data on tribal population leaves much to be desired [Davis 1951: 188]. While commenting on much lower share of people adhering to tribal religions in 1961 vis-a-vis that in 1931, Schwartzberg (1978, 233) attributes this decline to “reclassification as Hindus of a large number of tribal people who would have been classified in 1931 as belonging” to “tribal” religions rather than Hinduism. Such a reclassification might not have been considered improper as leading Indian scholars like Ghurye (1963, 19) had emphasised that tribal people “are in reality backward Hindus”. Whether the reclassification of tribal people as Hindus was due to lack of sharp boundary between animism and Hinduism or it was a manifestation of ongoing assimilative and homogenising urge of the ruling class in the country deserves an independent research investigation. However, looking at the way the three separate religions, i e, Sikhism, Buddhism and Jainism have been treated as part of the Hindu fold under the Article 25 of the Indian Constitution, one is within limits to conclude that the “reclassification” occurred due to assimilational urge of the ruling class, which, in line with religious composition of the country is preponderantly from the Hindu fold. There was a need as well as a demand for including a question on OBCs in the 2001 Census. But the then BJP-led central government of the country at the time of 2001 Census did not permit its inclusion as it did not go in line with its long-term political calculus [Bhagat 2003: 687]. Contrary to the earlier held view, the ruling party of the country (BJP) at that time seems to have come to believe that data along more caste lines also carry the potential for weakening/diluting what is now being called the political mainstream of the country. In contradistinction, its new assessment seems to hold that data on religion not only help to weld together and strengthen the mainstream, it also could be used, as is commonly done, for scapegoating some of the religious group(s) in order to enhance the vote bank of the ruling party.

Economic and Political Weekly

January 20, 2007

The release of census data on religion after every census invariably raises communal and political temperatures in India. The latest Census of 2001 was no exception in this regard. A number of important leaders of the Sangh Parivar were condemning the “explosive” growth rate of the Muslims, which was solely attributed to higher Muslim fertility rate. These leaders simply forgot to keep view of the fact that higher growth rate of Muslim population did not “automatically imply unbridled fertility” as lower Muslim mortality rate also played an important part in this connection [Basu 2004: 4295]. It was virtually ignored that Muslim fertility rate has come down significantly during recent years “following nearly the same pace of transition as that of Hindus [James and Nair 2005: 382]. This underlines the huge potential of socio-political misuse of religion data by vested interests. Besides, in some cases, the “illicit dramatisation of misrepresented statistics today” may be used as a justification for ethnic cleansing tomorrow [Jayaraj and Subramanian 2004: 4297]. Under this type of mindset, it becomes difficult to accept that the quality/accuracy of census data would remain uncompromised, particularly concerning politically sensitive issues like religion and language. Similarly, quality of data also come to suffer due to the lack of work commitment on the part of quite a good share of census enumerators, particularly those who feel themselves being compelled to do this task [Bose 2000: 1433].

Census and Caste Right from the beginning of the census in India in 1871, there have been strong views for and against collection of data on castes [Maheshwari 1996: 105; Deshpande and Sundar 1998: 2158; Burman 1998: 3179; Smith 2000: 5; Sundar 2000: 116-117; Upadhya 2002: 50]. The support for or opposition to census data by caste transcends Left-Right dichotomy as both “Left liberals and upper-caste Hindu Rightists” could be seen joining together on both sides of the fence [Sundar 2000: 123]. Those who support inclusion of caste for census purposes argue that since caste is an inseparable feature of India’s social life, it deserves to be reflected in the census returns. The opponents believe that caste data would work to erode the imagined solidarity of Hindu society. It is important to note that it is mainly the people from upper castes who are against census by caste, while those from lower castes invariably welcome such an effort [Sundar ibid: 115]. Caste data were also collected for all the religious groups, including the Christians, in the 1871 Census. However, as the protestant missionaries were opposed to inclusion of caste for the Christians, this practice was discontinued for them at the 1881 Census [Oddie 1981: 137]. Similarly, as the Muslim League was ideologically “against the mention of ‘caste’ among the Muslims”, the list of caste for the Muslims gradually got shortened after 1921 [Maheshwari 1996: 113]. Consequently, there was no mention of any caste among the Muslims for the 1941 Census. The benefits of caste reservation in the legislatures, government jobs and for admission to educational institutions were initially given to the Hindus only when the reservation policy was first introduced in 1950. The Sikhs were brought under the purview of this policy in 1954 and the Buddhists much later in 1990 [Oommen 2001: 228]. In other words, the government’s reservation policy for weaker castes has worked to perpetuate caste system not only among the Hindus, but also among the people of other two religions, i e, Sikhism and Buddhism which, in principle, are basically against the caste system as found in

245

Hinduism. Thus, by diluting opposition to caste system among these two religious communities, the reservation policy for and the census count of scheduled castes as a category has contributed toward homogenisation, or one could say Hinduisation, of India’s religious minorities which is in line with the political agenda of the country. The purpose of constitutional safeguards for the scheduled castes and scheduled tribes in terms of reservations in various fields was both economic and politico-religious. The explicit objective, no doubt, was to help these marginalised people to come out of their age-old socio-economic deprivation. The other equally important purpose of the reservation policy was to discourage them from converting to other religions. The fact that the benefits of the reservation policy were initially limited to the Hindus only points out that this seemingly pro-poor policy had religious underpinnings. It is notable that in the Indian setting, people of almost all religions are marked by caste divisions notwithstanding the fact that non-Hindu religions do not admit of; at least in principle, the caste system. Had the objective of the reservation policy for scheduled castes been solely economic, it could have easily been framed on economic criteria which would have rightly covered weaker sections from amongst all the religious communities, including the Hindus, in the country. At the time of the introduction of reservation policy for scheduled castes in 1950, the Sikhs had a difficult choice to make. Keeping in line with negation of caste system in Sikhism, if they had not accepted the prevalence of caste system among them, then it was very likely that a large share of Sikh scheduled castes would have opted for the Hindu fold to get the much needed socio-economic benefits of reservation policy. On the other hand, if they admit of the caste system, as they did quite quickly, it would not only go against the basic tenets of Sikhism, but would also push Sikhism closer to Hinduism as was planned and expected by the country’s mainstream leaders. Interestingly, it is now the Muslims and Christians only, who are not covered by the caste reservation policy notwithstanding the fact that a very large share of the followers of these two religions are the converts from weaker castes in India. No wonder, the pace of upward mobility in these strata among the Christians and the Muslims has been quite indifferent and the pace of conversions to these religions has also come down sharply, virtually to naught, in recent decades. Owing to the reservation policy for scheduled castes, the Muslims and the Christians could be said to be in a problematic position. If they accept the caste system to benefit from the reservation policy, they would not only be going against the basic principles of their religions, they would also be accepting a system, which is a hallmark of Hinduism. On the other hand, if they do not opt for it, they would not be able to make available the much-needed and well-deserved benefits of the reservation policy to the weaker sections/castes amongst them. Undoubtedly, the publication of data by scheduled castes as a single category has been helpful in strengthening the Bahujan Samaj Party (BSP) and some other caste-based parties, which mainly articulate the interests of weaker castes of India. Though such parties have eaten into the vote banks of other dominant parties like the Congress and the BJP, these operate essentially within the overall framework of Hinduism, albeit of nonbrahminical hue [Bhambhri 1999: 2620]. Similarly, the rise of caste-based political parties works to enhance horizontal integration of various castes [Smith 2000:5]. Such a process also helps in the integration of this highly diverse country, which

246

expectedly, has been among the top agenda of the country’s rulers. On the other hand, religion-based political parties are often coldshouldered under the majoritarian/mainstream view, as these are perceived to go against the process of homogenisation. The publication of data by scheduled castes and relative paucity of the same by religion for five decennial censuses during the postindependence period need to be viewed in the above context.

Some Evidence from Punjab If one looks at the growth rate of different religious communities, one is struck with the fact that the Hindus have experienced a consistent decline in population in India (Table 1). On the other hand, the minority religious communities like the Muslims, the Sikhs and the Buddhists have improved notably their share in the country’s population. Such data convey a wrong message that the minorities breed faster as compared to the majority people of the country, and hence may allegedly outnumber the majority Hindu community in the long run. The presence of data on religious communities’ size of population alone when taken in an undigested form is sure to raise a community consciousness in various ways which often suits the interests of the ruling classes in plural countries. The same thing is happening in the case of India too. Similarly, even data on total population in certain parts of the country are sometimes under-reported due mainly to political reasons. For instance, in case of Bihar, Uttar Pradesh and Rajasthan, the 1981 census estimates of population “were actually less than those indicated by the housing census of the previous year” [Dyson 1981: 194]. Sometimes, even the veracity of census data is quite suspect. The Census of India 1981 [Registrar General and Census Commissioner, India 1983: 39-40] revealed that the Sikhs had lower total marital fertility rate than that of the Hindus in Punjab, where 78 per cent of the Sikh in India resided at the time of that census (78.52 per cent in 1991 and 75.94 per cent in 2001). Interestingly, in spite of lower fertility as compared to that of the Hindus, the Sikhs’ proportion had gone up to the country’s population during 1971-1981 and 1981-1991 as they had recorded relatively high growth data than that of the Hindus. As the Sikhs have been registering an excess of emigration over immigration during the last five decades, relatively high growth of these people was possible only if they had higher fertility than that of the Hindus and/or there was notable incidence of religious conversion to Sikhism from the Hindu fold. As the Sikhs had distinctly lower total marital fertility rate and conversions from Hinduism to Sikhism have been, if at all, simply negligible in the post-1947 period, it becomes difficult to explain Table 1: Proportion of Religious Communities to Total Population of India*, 1961-2001

Hindus Muslims Christians Sikhs Buddhists Jains Others

1961

1971

1981

1991

2001

84.4 9.9 2.4 1.8 0.7 0.5 0.3

83.5 10.4 2.6 1.9 0.7 0.5 0.4

83.1 10.9 2.5 2.0 0.7 0.5 0.4

82.4 11.7 2.3 2.0 0.8 0.4 0.4

81.4 12.4 2.3 1.9 0.8 0.4 0.7

Note: * Excludes Jammu and Kashmir and Assam for all decades from 1961 to 2001. Source: Registrar General and Census Commissioner, India (2004): Census of India 2001, The First Report on Religion Data, xxviii.

Economic and Political Weekly

January 20, 2007

higher growth rate of the Sikhs vis-a-vis the Hindus in India. The only plausible explanation of this could be that the figures relating the Sikh population got somewhat inflated for political reasons. A quite rapid decline in the growth rate of Sikhs from 25.5 per cent during 1981-91 to 16.9 per cent during 1991-2001 could partly be attributed to over-recording of their population at the 1991 Census. Data on Sant Nirankaris and Radhaswamis: These two sects have considerable focus of their activities on Punjab-Haryana-Delhi area. While addressing their followers, the leaders of these sects quite frequently quote from the Adi Granth, the Holy Scripture of the Sikhs. Similarly, like the Sikhs, the chiefs of the Sant Nirankaris and the Radhaswamis keep their hair unshorn. In other words there is seemingly notable resemblance between the Sikh preachers on the one hand and those of the Sant Nirankaris and the Radhaswamis on the other. However, the concept of living guru among both the sects as well as the basic thrusts of their religious discourses go against the very principles of Sikhism. Until their bloody confrontation with the Sikhs in 1978, the Sant Nirankaris were growing quite fast in Punjab and its adjoining areas. However, for the last two decades, it is now the Radhaswamis who are spreading rapidly as manifested in large congregation places (‘satsang bhawans’) which have mushroomed in all the urban places as well as in many of the big villages in different parts of Punjab during this period. The veracity of data regarding religious sects and persuasions is not only under-enumerated, but is strongly coloured by political considerations. For instance, enumeration of the populations of the Sant Nirankaris and the Radhaswamis has been marked by notable undercount since 1971 Census. As per the 1971 Census of India, there were only 34 Nirankaris in Punjab and all of them were shown having been enumerated in Jalandhar district [Registrar General and Census Commissioner of India 1976: 24]. This figure is totally unbelievable as the Sant Nirankaris had their large satsang bhawans in most of the urban centres of Punjab in that year. All this could not have been done by and for these 34 adherents of this persuasion. Besides, the unrealistic nature of the figure was manifest from another angle also. In 1978, i e, just seven years after the 1971 Census, the Sant Nirankaris were engaged in a bloody confrontation with the Sikhs in Amritsar city, the main religious centre of the Sikhs who, with a population of 8,159,972, accounted for 60.22 per cent of Punjab’s population in 1971. Thus, merely 34 persons and their well-wishers could not have been in a position to come out with a militant challenge to the majority religious community of Punjab, and that too in its main religious city. In other words, the data furnished by the 1971 Census regarding the number of Sant Nirankaris were much below the actual figures. The situation in this regard was no better even at the 1991 Census, which recorded only 551 Sant Nirankaris in whole of India and merely 197 in Punjab (Table 2). In other words, there were approximately 100 families belonging to this sect in the country. This figure is certainly not in tune with millions of rupees worth of property and other assets in the name of Sant Nirankari Bhawans in different urban centres, particularly those in Punjab, Haryana and Delhi. The same could be said about census enumeration of the Radhaswamis in Punjab. As per the 1971 Census, there were only 451 Radhaswamis in Punjab [Registrar General and Census Commissioner of India 1976: 24]. Of them, only four were recorded in Amritsar district notwithstanding the fact that a place

Economic and Political Weekly

January 20, 2007

called Beas, being the headquarters of this sect, had a notable concentration of the Radhaswamis. The number of adherents of this sect has been given as 3,788 in 1981 [Registrar General and Census Commissioner of India 1984: 1499] which declined to 2,481 in 1991 [Registrar General and Census Commissioner of India 1996: 403]. notwithstanding the fact that followers of this sect had increased notably during this decade. If one looks at huge property in the name and Radhaswami satsang bhawans at prime locations, worth millions of rupees in some cases, primarily along major routes in almost all the urban centres and in many of the big villages of Punjab, one feels compelled to disbelieve these population figures. Significantly, the census gives figures for the Hindu Radhaswamis only while it is silent concerning Radhaswamis from the Sikh community despite the fact that spread of this sect has been greater among the Sikhs. It could not be accepted that there was no Radhaswami family from the Sikh fold at the 1991 Census in Punjab. This view is further confirmed by the fact that the 1981 and 1991 Censuses recorded Radhaswamis in all the districts of Punjab save in Amritsar district. Incidentally, the headquarters of this religious sect are at a place named Beas in Amritsar where a notable Radhaswami population has been in residence since long. Such a glaring lapse in not recording any Radhaswami population in their main area of concentration could not be attributed to a mere oversight error. The over under-count of an ethnic group could take place due to the role of following four factors which might work singly or in a particular combination: (i) the simple carelessness or indifference on the part of the enumerators towards census-count; (ii) the absence of clear differentiation between an ethnic group and the basal population from which the groups has grown out; (iii) the strategic effort on the part of the ethnic group toward over- or under-count of its population; and (iv) an intentional over- or under-recording at the instance of the state. Obviously, the last two factors are directly of political nature. In case of census under-count of the Radhaswamis, the indifferent or careless attitude of census enumerators might have certainly made quite perceptible effect in this regard as could also be noticed concerning under-enumeration of many other sects/sub-groups of other religions [Bose 2000:1433]. However, this factor alone could not account for total absence of Radhaswami sect in Amritsar district in 1991, wherein its headquarters are locatd at the settlement named Beas, which is very well known as Radhaswami Dera, in Punjab. Similarly, lack of “sharp difference” between Radhaswamis on the one hand and the basal populations of the Sikhs and the Hindus on the other might Table 2: Population of Sant Nirankaris in India, 1991

India Punjab Delhi Maharashtra Himachal Pradesh Rajasthan Haryana Madhya Pradesh West Bengal Chandigarh Uttar Pradesh Orissa

Total

Rural

Urban

551 197 120 44 37 36 35 26 26 22 5 3

148 70 23 33 15 6 1 1 -

403 127 120 21 4 21 29 26 25 22 5 3

Source: Registrar General and Census Commissioner, India (1996): Census of India 1991, India, Religion, Part IV-B(ii), Table C-9, 272-275.

247

have made only a marginal difference in this connection, and that too in case of fresh entrants into this sect. Thus, the remaining two factors, i e, wilful wrong returns by the respondents as well as the role of state policy have been primarily responsible for unusually low census-count of Radhaswamis in Punjab, and particularly their total absence in Amritsar district. Similarly, published data from the 1981 and 1991 Censuses reveal that there were no Radhaswamis among the Sikhs notwithstanding the fact that successive chiefs of this sect came from the Sikh fold. Whether the absence of this sect among the Sikhs is attributed to false returns by the respondents or to state policy, it necessarily carries strong political implications. The same applies to unusually low census returns of the Sant Nirankaris and the Kukas in Punjab. In actual terms, the population of the Radhaswamis is much larger. But it suits the country’s ruling class to underplay their population size. It is helpful to the ruling class politically as it does not disturb the Sikh community whose base is systematically being eroded in Punjab through the spread of such sects. The Sikhs have been emphasising their distinctive political discourse during the post-independence period, any erosion of their demographic and, hence, political base would certainly be to the liking of the country’s rulers. Hence, the non-recording of data on the Radhaswamis who come from the Sikh fold serves well the interests of the major political parties of the country. Interestingly, the Sikh political leadership itself also seems to find non-recording of this type of data in its own interest since it helps avoid not only the resentment among the Sikh masses in this connection, but it also allows the obtaining Sikh political leadership to stay on the favourable side of India. Similarly, as per the 1991 Census, the Sant Nirankaris have been included in the category “other Religions and Persuations” suggesting that this group has been considered as a distinct religion or persuasion. In contrast, the Radhaswamis, with much larger population as given in the census, have been clubbed with Hindu religion. This differential categorisation of these two communities, which are almost identical in their preaching and objectives, is another subtle manifestation of obtaining politics of census data in the country. It may be pertinent to point out that the Sikhs’ socio-political and religious identity in the Indian Punjab has been one of the major obstacles, at least till the early 1990s, for the homogenising process launched by the Indian state since the end of the British rule in 1947. In order to blunt the thrust of the distinct Sikh discourse, every effort has been made by India, both in overt and covert manner, to create fissures and variants among the Sikhs. Among others, the political patronage to the Sant Nirankaris till about 1978, and subsequently covert encouragement to the Radhaswamis and to some other similar sects since early 1980s could be cited as two major strategies adopted by India in this connection. The main objective remains to ethnicise and fragment the Sikhs so that it becomes easier for the forces of homogenisation to dilute and absorb the difference as represented by the Sikhs, particularly in Punjab. The tactics adopted in terms of under-reporting of data on new religious persuasions as well as their characteristic categorisation in terms of various sects in Punjab is indicative of the role of political factors. In spite of the fact that the numbers of adherents of these sects, particularly the Radhaswamis, have been continuously on the rise in recent decades, these are grossly underplayed in census returns so that the Sikhs do not become adequately aware

248

of the erosion of their population base. In contradistinction, the population figures for the Sikhs seem to be somewhat inflated in censuses from 1961 to 1991. Similar processes could also not be ruled out completely in areas having notable concentration of tribal and other minority people.

Summing up The collection and publication of population data by the census organisations is often characterised by strong political overtones. Apart from the enumeration of various demographic facts, census taking enhances a characteristic visibility of certain aspects/ groups of population in tune with the reigning political narrative at a given point of time. In most of the developing plural countries, like India, the religious and ethnic minorities are often given much less attention in census details as compared to even the sub-groups of the majority population of the country. Gradually, such minorities get hidden from the view or are projected as mere variants of the majority population toward which the design and detail of population census data make their own contribution in a significant manner. The categories as well as details of Indian census data have been closely connected with the political agendas of the country’s rulers during the British as well as post-British periods. The curtailment of census categories on religion, tribes and also on castes during the post-independence period is mainly attributable to the introduction of strongly homogenising political discourse by the Indian state. Interestingly, whereas the British Indian census contributed to locate precise boundaries between once “fuzzy” communities, the post-British census’ important objectives have been to re-demarcate many of these boundaries, and to reintroduce some “fuzziness” in this connection. It is important to note that while the Hindu majority population is generally opposed to the collection of detailed data by religion, the religious minorities are mostly in favour. Similarly, upper-caste Hindus are often against census data by castes, while lower castes mostly welcome it as it would help reinforce their social identity which is becoming increasingly important for enhancing political strength in the evolving situation. The publication of religion data from the 2001 Census in significantly greater details is a highly welcome step, which deserves a separate study regarding its socio-political implications. EPW Email: [email protected]

References Ananth, J K (2001): ‘The Political Economy of Communalism: Some Observations on the Contemporary Political Discourse’, Social Scientist, 29 (7-8), 48-67. Banerjee, Sumanta (2003): ‘Bengali Left: From Pink to Saffron’, Economic and Political Weekly, 38(9), March 1, 864-65. Barrier, N G (1981): ‘Introduction’ in N G Barrier (ed), The Census in British India, Manohar, Delhi, v-xiv. Basu, Alaka, M (2004): ‘The Squable that Never Ends: Religion and Fertility’, Economic and Political Weekly, 39 (39), September 25. Bava, U S (2003): ‘Dynamic India: Imagined and Real Review Article’, South Asian Survey, 10(2), 275-80. Bhagat, R B (2001): ‘Census and the Construction of Communalism in India’, Economic and Political Weekly, 36, (46 and 47), November 24. – (2003): ‘Role of Census in Racial and Ethnic Construction: US, British and Indian Censuses’, Economic and Political Weekly, 38(8), February 22.

Economic and Political Weekly

January 20, 2007

– (2004): ‘Fact and Fiction’, Economic and Political Weekly, 39(39), September 25. Bhambhri, C P (1999): ‘Dialectics of Caste and Casteism’, Economic and Political Weekly, 34 (36), September 4. Bose, A, D B Gupta and G Raichaudhury (eds) (1977): Population Statistics in India, Vikas Publishing House, New Delhi. Bose, A (1991): Population of India: 1991 Census Results and Methodology, B R Publishing Corporation, Delhi. – (2000): ‘Beyond a Headcount: Preparing for Census of 2001’, EPW, April 22. – (2005): ‘Beyond Hindu-Muslim Growth Rates: Understanding SocioEconomic Reality’, Economic and Political Weekly, 40(5), January 29. Brass, P R (1992): The Politics of India since Independence, Cambridge University Press, New York. Bulmer, M (1979): ‘Parliament and the British Census since 1920’ in M Bulmer (ed), Census Surveys and Privacy, Macmillan Press, London. Burman, B K R (1998): ‘Backward Classes and the Census’, Economic and Political Weekly, 33(50), December 12. Census of Ethiopia (1998): The 1994 Population and Housing Census of Ethiopia, Vol I, Statistical Report. Christopher, A J (2002): ‘‘To Define the Indefinable’: Population Classification and the Census in South Africa’, Area, 34(4), 401-08. Cohn, Bernard S (1987): ‘The Census, Social Structure and Objectification in South Asia’ in Bernard S Cohn, An Anthropologist among the Historians and Other Essays, Oxford University Press, New Delhi, 224-54. Das, A N (1992): India Invented: A Nation-in-the Making, Manohar, New Delhi. Davis, K (1951): The Population of India and Pakistan, Princeton University Press, Princeton. Deshpande, S and N Sundar (1998): ‘Caste and the Census’, Economic and Political Weekly, 33(32), August 8. Dyson, T (1981): ‘India’s Demographic Transition: An Appraisal of the Provisional 1981 Census Results’ in K Srinivasan and S Mukerji (eds), Dynamics of Population and Family Welfare 1981, Himalaya Publishing House, Bombay, 189-204. Escobar, A (1984): ‘Discourse and Power in Development: Michel Foucault and the Relevance of His Work to the Third World’, Alternatives, 10, 377-400. – (1997): ‘Planning’ in W Sachs (ed), The Development Dictionary: A Guide to Knowledge, Orient Longman, New Delhi, 176-194. Foucault, M (1980): Power/Knowledge, C Gordon (ed), Translated by Gordon, C et al, Pantheon Books, New York. Gait, E A (1913): Census of India 1911, India, Part I – Report, Vol I, Superintendent, Government Printing, Calcutta. Ghurye, G S (1963): The Scheduled Tribes, Popular Prakashan, Bombay. Hutton, J H (1933): Census of India, 1931, Vol I, India, Part II – Imperial Tables, Manager of Publications, Delhi. Inden, R (1990): Imagining India, Basil Blackwell, Oxford. Jaffrelot, C (1996): The Hindu Nationalist Movement in India, Columbia University Press, New York. James, K S and S B Nair (2005): ‘Accelerated Decline in Fertility in India since the 1980s: Trends among Hindus and Muslims’, Economic and Political Weekly, 40(5), January 29. Jayaraj, D and S Subramanian (2004): ‘Manufacturing Hysteria: On Census-Inspired Nationalism’, Economic and Political Weekly, 39(39), September 25. Jenkins, R (2000): ‘Categorisation : Identity, Social Process and Epistemology’, Current Sociology, 48 (3), 7-25. Jones, K W (1981): ‘Religious Identity and the Indian Census’ in N G Barrier (ed), The Census in British India, Manohar, Delhi, 73-101. Kaviraj, S (1993): ‘The Imaginary Institution of India’ in P Chatterjee and G Pandey (eds), Subaltern Studies VII, Oxford University Press, Delhi, 1-39. Khilnani, S (1997): The Idea of India, Penguin Books, London. Kulkarni, S (1991): ‘Distortion of Census Data on Scheduled Tribes’, Economic and Political Weekly, 25(5), February 2. Maheshwari, S (1996): The Census Administration under the Raj and After, Concept Publishing Company, New Delhi. Mawdsley, E (2002): ‘Regionalism, Decentralisation and Politics State Reorganisation in Contemporary India’ in R W Bradnock and G Williams (eds), South Asia in a Globalising World, Pearson Education Limited, Harlow, Essex, 122-143.

Economic and Political Weekly

January 20, 2007

Mohanty, S P (1996): ‘Significance of Census Data in the Context of Contemporary Indian Society’ in S P Mohanty and A R Momin (eds), Census as Social Document, Rawat Publications, Jaipur, 162-167. Momin, A R (2002): ‘Celebrating Diversity: Review Article’, Sociological Bulletin, 51(1), 101-115. Nakarada, R (1983): ‘Marxist Praxis: Politics of Power a Way of Life’, Alternatives, 9, 377-385. Oddie, G A (1981): ‘Christians in the Census: Tanjore and Trichinopoly Districts 1871-1901’ in N G Barrier (ed), The Census in British India, Manohar, New Delhi, pp 119-49. Oommen, T K (2001): ‘Civil Society: Religion, Caste and Language in India’, Sociological Bulletin, 50(2), 219-35. Pandey, G (1998): ‘In Defence of Fragment: Writing about Hindu-Muslim Riots Today’ in R Guha (ed), A Subaltern Studies Reader, 1986-1995, Oxford University Press, Delhi, 1-33. Patel, V B (1950): ‘Opening Address to Census Conference February 23March 1’ in The Registrar General India Census of India (1973): Indian Census through a Hundred Years, Part I (Chapters I & II), Census Centenary Monograph No 2, The Manager of Publications, Delhi, 265-66. Registrar General and Census Commissioner, India (1972): Census of India 1971, India, Religion, Paper 2 of 1972, The Manager of Publications, Delhi. – (1976): Census of India 1971, Punjab, Social and Cultural Tables, Part II-C (i) and Part V-A, The Controller of Publications, Delhi. – (1977): Census of India 1971, India, Social and Cultural Tables, Part II-C (i), The Controller of Publications, Delhi. – (1983): Census of India 1981, Punjab, Report and Tables Based on 5 Per Cent Sample Data, Part-II – Special, The Controller of Publications, Delhi, 39-40. – (1984): Census of India 1981, India, Household Population by Religion of Head of Household, Paper 4 of 1984, The Controller of Publications, Delhi. – (1996): Census of India 1991, India, Religion, Part IV-B(ii), Table C-9, The Controller of Publications, Delhi. Said, Edward W (1994): ‘Identity, Authority and Freedom: The Potentate and the Traveller’, Boundary 2, 21 (3), 1-18. Schwartzberg, J E (1978): A Historical Atlas of South Asia, The University of Chicago Press, Chicago. – (1981): ‘Sources and Types of Census Error’ in N G Barrier (ed), The Census in British India, Manohar, Delhi, 41-60. Singh, K S (1996): ‘Census and Ethnography’ in S P Mohanty and A R Momin (eds), Census As Social Document, Rawat Publications, Jaipur, 138-46. Singh, P (1999): The Sikhs, Harper Collins Publishers India, London. Smith, R S (2000): ‘Between Local Tax and Global Statistics: The Census as Local Record’, Contributions to Indian Sociology, (NS), 34 (I), 1-35. Srivastava, H C (1996): ‘Ethnography in Indian Census’ in S P Mohanty and A R Momin (eds), Census As Social Document, Rawat Publications, Jaipur, 206-220. State Committee of the Russian Federation on Statistics (1998): Population of Russia: 1897-1997, Statistical Abstract, Moscow. Sundar, N (2000): ‘Caste as Census Category: Implications for Sociology’, Current Sociology, 48 (3), 111-126. Upadhya, C (2002): ‘Hindu Nationalist Sociology’ in G S Ghurye (ed), Sociological Bulletin, 51(1), 28-58.

Economic and Political Weekly Available from Star News Agency Mahendra Chambers Magazine Market, 146, D N Road Mumbai - 400 001

249