In the first decade of the twentieth century, the autocratic power of the British Raj reached its height, but it also started to decline, especially after the partition of Bengal in 1905, which gave a decisive impetus to the Indian nationalist movement. Four years later, the Morley-Minto reforms became law. They were designed to make concessions to ‘moderate’ Congressmen, whose support was concentrated in the predominantly Hindu, educated, urban middle class, but also to counterbalance them with special provisions for Muslims and wealthy landlords opposed to the nationalists. In the electoral system for the new legislative councils, the basic principle was that the middle class was represented by members elected from open, territorial constituencies, whereas Muslims and landlords voted in separate electorates. The Morley-Minto reforms therefore introduced Muslim ‘communal’ representation, which meant that the size of India's Muslim minority – and conversely the Hindu majority – had clear political consequences for the first time.
In 1919, the Montagu-Chelmsford reforms became law. Implemented in India in 1921, these reforms introduced the ‘dyarchy’ system, which devolved limited governmental responsibility to Indian ministers elected to provincial legislative councils. Compared with 1909, the franchise was considerably expanded and so, too, was the scope of communal representation, which was extended for Muslims and also introduced for other groups, such as Sikhs in the Punjab. In the new system, provincial politics became increasingly organised along caste lines, especially in Madras and Bombay by the opposition between Brahmans and Non-Brahmans. The last major constitutional reform before Independence was the 1935 Government of India Act, which abolished dyarchy and replaced it with a federal system that placed provincial government in the hands of ministries responsible to enlarged electorates. Communal representation for Muslims and other religious minorities remained an important feature, and so did the earlier patterns of caste-based politics, but a crucial innovation in 1935 was that reserved seats – though not separate electorates – were allotted to the Scheduled Castes, as the Untouchables were now officially designated. All in all, therefore, politicised caste and ‘casteism’, as well as religious communalism, were strengthened during the inter-war period.
The systematic, ‘scientific’ anthropology of India began in the late nineteenth century and developed in close connection with the decennial censuses, the first of which was held in 1871–2. ‘Traditional’ Indian society – the antithetical ‘other’ of modern European society – was, the British believed, primarily made up of castes with a tribal periphery, divided into separate religious communities. Information about castes, seen as the most important social groups, could be collected through surveys, especially censuses, and – as Bernard Cohn explained – the “census-based view of caste . . . saw the system as one of separate castes and their customs”, which had to be counted and classified.Footnote 2 Cohn's pioneering work on colonial anthropological knowledge in India has been variously developed by subsequent scholars. They have collectively shown that by counting, classifying and describing castes and religious communities as separate, reified, ‘substantialised’ or ‘ethnicised’ groups (in more recent terminology), the censuses and colonial anthropology significantly encouraged the development of politicised caste and religious communalism during the colonial period. Equally or more important, of course, were British policies predicated on the separateness of groups, especially Hindus and Muslims, which were often intended to ‘divide and rule’.Footnote 3
This thesis about colonial anthropology and the colonial state is basically correct, I believe, and so is Nicholas Dirks's proposition that in India anthropology became “the principal colonial modality of knowledge and rule”, so that, by the late nineteenth century, “the colonial state . . . can be characterised as the ethnographic state”.Footnote 4 But in at least two major respects the thesis needs qualification. In another article, I have examined the role of Denzil Ibbetson and Herbert Risley, the two leading official anthropologists of the period, in the partition of Bengal and the Morley-Minto reforms, when they held high office as, respectively, the Home Member of the Viceroy's Council and the Secretary of the Home Department between 1902 and 1910.Footnote 5 Very little of Ibbetson's and Risley's policy advice to the Viceroys was in fact shaped by their anthropological understanding of Indian society. That was mainly because the two most critical problems facing the government were the growing strength of the nationalist movement, whose support was concentrated in the educated, urban middle class, and Muslim demands for separate representation. But colonial anthropology, which focused entirely on ‘traditional’, mainly rural, society, provided no information about the middle class; furthermore, all the ethnographic data on popular religion showed that Hindus and Muslims did not actually belong to two distinct, bounded communities. To put it bluntly, anthropology was useless on the first issue and gave the wrong answers on the second. This situation did not improve later and even for defining the Untouchable castes, which mattered for the 1935 Act, anthropology could not solve the problem at hand. The colonial state, in the Cohnian thesis, is a complex structure that encompasses more than just the government itself. Nonetheless, the government is plainly a crucial part of the state. It is therefore important for analysis of the colonial Indian state – as this article will show – that from around 1909 to 1939 ethnography and anthropology became progressively less salient for high policy in the imperial government, though not necessarily for decision making at lower levels of the administrative and judicial system.
The second flaw in the thesis about colonial anthropology pertains to its academic development, rather than its political aspects. Because Cohn and other scholars have concentrated on the nineteenth century, ending with the 1901 census, they have overlooked the reduced attention paid to collecting data about caste from 1911, as well as the significant advances made in understanding the caste system when the problems of reification were recognised and partly overcome. The anthropological work of three officials, who have been largely ignored by historians of colonial knowledge and forgotten by contemporary social scientists, was particularly valuable. The first of them, Edward Albert Gait (1863-1950), was the Superintendent for Assam at the 1891 census, the Superintendent for Bengal and afterwards the Commissioner in overall charge at the 1901 census, and the Commissioner at the 1911 census. The other two were Edward Arthur Henry Blunt (1877-1941) and Lewis Sydney Steward O'Malley (1874-1941), who were the 1911 census Superintendents for the United Provinces and Bengal, respectively; both men later wrote important books on caste. Gait and Blunt, as senior officials, also played a significant part in the political developments of their day, which I shall discuss as well.
Colonial Anthropology and the Censuses of 1881, 1891 and 1901
Census officials and anthropologists in British India – including Ibbetson, Risley, Gait, Blunt and O'Malley – were almost all members of the elite Indian Civil Service, the ICS, who were known in their day simply as ‘civilians’. In the period covered by the 1881, 1891 and 1901 censuses, Ibbetson and Risley made the most notable contributions to the ethnography and theory of caste, and to the promotion of anthropological enquiry.Footnote 6 Ibbetson's 1881 census report on the Punjab, the first major publication on caste emerging from the censuses, included an ethnographic description of the province's castes and tribes, as well as his theory that castes are fundamentally occupational groups, a product of the evolution of the division of labour.Footnote 7 In 1885, the Government of Bengal deputed Risley to carry out the research that he later published in a handbook, The Tribes and Castes of Bengal.Footnote 8 In 1901, Risley was appointed Honorary Director of Ethnography for India, responsible for a project to complete tribes and castes handbooks for every province of British India. He was also the 1901 census Commissioner, until he was promoted and replaced by Gait; Risley's chapter on caste in the census report became the basis for his book, The People of India, published in 1908.Footnote 9 Risley, who criticised the occupational theory, argued that the hierarchical caste system was ultimately based on distinctions of race, primarily between the ‘advanced’ Aryans and indigenous, ‘primitive’ Dravidians. He was interested, too, in the ‘social precedence’ or ranking of castes, and he believed that anthropometric bodily measurements could prove that a correspondence existed between caste rank and physical features. Risley's emphasis on ranking greatly influenced the subsequent study of caste, whereas his racial theory – which failed to convince most of his colleagues – has been completely discredited.Footnote 10
In the 1891 and 1901 censuses in particular, huge amounts of data on caste were collected throughout British India, as well as in most princely states, and they were tabulated to produce long lists of castes and subcastes, with population numbers. The reports also contained ethnographic and theoretical accounts of caste, as well as tribe and race. All the theory is now outmoded, but substantial progress was made in documenting and understanding the caste system from 1881 to 1901.
All of these census reports also included population figures for different religious groups. The 1881, 1891 and 1901 ones contained a lot of material about religion as well: usually a historical survey of Hinduism and Islam (and sometimes other religions), a general account of beliefs and practices at both elite and popular levels, and information on old and new ‘sects’ and reformist movements. The section on Hinduism normally accompanied one on the ‘animism’ characteristic of ‘primitive’ tribes, but also part of the religion of low-status, marginal, ‘semi-Hinduised’ groups. In the reports, three points were emphasised repeatedly. First, popular religion is as much ‘social’ as ‘religious’ in a credal sense. Secondly, among the great majority of ordinary, uneducated villagers – irrespective of whether they are nominally Hindus, Muslims, or followers of another religion – popular religion is an eclectic melange of beliefs and practices, typically combining Hinduism with Islam or another faith, as well as animism. Thirdly, whereas Islam or Christianity can in principle be defined by their beliefs and scriptures, Hinduism is fundamentally indefinable, because it loosely encompasses myriad traits, from the high Brahmanical to the low animistic.
In 1891, Alfred Lyall described Hinduism as “a conglomerate of rude worship and high liturgies” and “superstitions and philosophies belonging to very different phases of society and mental culture”.Footnote 11 Lyall was regularly quoted by official anthropologists, some of whom collected detailed ethnographic material, but in general they made less progress in understanding religion than caste. That was partly because the evidence was persistently interpreted with reference to evolutionary theories of religion and often unproductive debates about whether, for instance, E.B. Tylor's influential concept of ‘animism’ – supposedly the most primitive religion – could be applied in the Indian case.Footnote 12 Consequently, the continuity between ‘Brahmanism’ and ‘animism’ – for example, the fact that ‘little’, village deities are commonly forms of ‘great’, Sanskritic ones – was rarely acknowledged. There was also little examination of the relationship between religious diversity and the social structure. To modern anthropologists, at least one facet of this relationship is almost self-evident, but W. Francis was actually exceptional in noticing that in Madras one index of caste rank was ‘adherence’ to Brahmanical ritual.Footnote 13 By contrast, the debates about caste between the rival occupational and racial theories, albeit evolutionist as well, did lead to better understanding of the contemporary caste system. The comprehension of Hinduism was probably impeded, too, by a kind of mental block among the British, whose Christian, usually Protestant, background tended to provoke bewilderment at polytheistic ‘idolatry’. Monotheistic Islam posed no such problem, but that in turn encouraged a false assumption that the religion of Indian Muslims was easy to grasp, together with a persistently oversimplified opposition between India's two main religions. O'Malley's later book on popular Hinduism was a useful contribution; he described Hinduism mostly sympathetically, unlike many colonial writers, and he also briefly recognised that there is no true separation between Hinduism and animism.Footnote 14 O'Malley saw too that “The higher we go in the social scale, the more does Brahmanical worship prevail”.Footnote 15 In general, though, the overall quality of writing about religion by official anthropologists was inferior to that on caste, which is mainly why I discuss its study here at greater length.
The Problem of Reification
In discussing the ‘census-based’ view of ‘separate castes and their customs’, Cohn pointed out that official anthropologists assumed that “a caste was a ‘thing’, an entity which was concrete and measurable”, with “definable characteristics – endogamy, commensality rules, fixed occupations, common ritual practices”.Footnote 16 In this positivistic reification of caste, there are two particularly crucial flaws. First, if castes are seen as separate things, the focus is on elements rather than relations, on substance rather than structure, so that comprehending the caste system as a system, either locally within a village or on a wider regional or pan-Indian level, becomes impossible. Secondly, reification precludes any proper cognisance either of what Indians themselves mean by jati or other indigenous terms that may be translated as ‘caste’, or of how their subjective perceptions are themselves constitutive of the institution. Gait, Blunt and O'Malley partly discerned these flaws of reification, although they did not of course express themselves in the language of modern social science. But Gait did emphasise the importance of indigenous perceptions, Blunt set out a proto-functionalist analysis of the system in one region, and O'Malley developed a relational understanding of caste as a pan-Indian system. These three civilians made significant progress in making the caste system more intelligible in ways that foreshadowed modern anthropologists and sociologists after Independence.
“Statistics accumulate and knowledge decays”: this was the epigram provocatively placed on the title page of the 1901 Punjab census report by H.A. Rose, who was unconvinced, too, by Risley's attempt to classify castes by “social precedence” in the census.Footnote 17 Rose was an unusually sceptical official anthropologist, but others shared his suspicions about the value of census data. Cohn is right that census reports in particular – but also tribes and castes handbooks and other publications – do portray castes, tribes and religious communities as separate, countable ‘things’. Yet the same literature also shows that reification was attained only after struggling with the errors and inconsistencies thrown up by the attempt to make messy data fit into neat categories. Nobody was more aware of these struggles than census superintendents themselves.
In his 1881 Punjab census report, Ibbetson discussed the errors in his data on castes and tribes in detail. Some problems, common to all censuses, were caused by confusing variations in the spelling of names and arithmetical mistakes, but there were also more significant ones. For example, it was difficult to translate the enumerators’ English schedule into the vernacular. In the Punjab, the schedule had three columns: for “the caste or race”, “the tribe proper”, and “the section of the caste”.Footnote 18 But qaum, for instance, could denote ‘religion’ or ‘caste’, and zat ‘caste’, ‘tribe’ or ‘clan’; moreover, confusion was compounded because Ibbetson had underestimated the variation in vernacular terms. It was then discovered that the completed schedules contained the names of thousands of ostensibly different castes, and still more subdivisions, so that there was time to check only the entries for castes (but not tribes) after tabulation. Ibbetson referred to scepticism about the census's accuracy among district officers, but still insisted that in the province as a whole, “the errors are probably insignificant, and hardly affect the general distribution of the population by caste”.Footnote 19 It is impossible to tell whether his satisfaction was well founded. On the other hand, Ibbetson clearly knew that Punjabis could readily name their own social groups, so that the problem lay in imprecise correspondence with the enumerators’ schedules. He accepted that errors could never be entirely eliminated in future censuses, but they could be reduced by simplifying the schedules and using a more systematic list, which Ibbetson intended to produce, but never actually did.Footnote 20
Ibbetson's discussion shows he was less persuaded by the fallacy that castes are things than many of the findings in his published report would suggest. In the early censuses, a dozen superintendents mentioned errors and reached conclusions similar to Ibbetson's. R. Burn, the Superintendent for the United Provinces at the 1901 census, was especially perceptive. Burn emphasised that Indians had “indefinite views” about the defining characteristics of castes and their subdivisions; like Ibbetson, he observed that vernacular terms such as jat (‘caste’) had variable referents, while conversely English terms such as ‘clan’ were not translatable by a single vernacular word. Rejecting the idea then being debated that the endogamous subcaste should be defined as the ‘caste proper’ for scientific enquiry, Burn argued that it is always better “to recognise as castes those endogamous groups which are considered as castes by the people themselves”, but “finality cannot be hoped for”, because new endogamous groups continually emerge and “public opinion as to what is a caste varies in different districts and at different times”.Footnote 21
Gait on Caste
Gait, like Burn, saw that how castes are defined necessarily depends on people's perceptions, so that they are never concrete things that can just be listed on census enumerators’ schedules. But Gait developed the critique of reification more fully and was probably led to do so by his 1901 census of Bengal, which was one of the most determined attempts made at any census to provide accurate data on each and every caste. According to Gait, the “more ignorant classes” have little idea about their own caste, so that they tell enumerators their occupations, subcastes, clans, titles and so on.Footnote 22 Partly for that reason and partly because so much regional variation existed within the province of Bengal, the caste return in the 1891 census had been defective. Gait's complaint about ignorant respondents was misplaced, but he was right that revising the 1891 list would improve enumeration and tabulation, and he eventually wrote an exceptionally detailed discussion of caste, accompanied by an exceptionally comprehensive list of castes and tribes, which probably contained fewer spurious entries and muddled names than any other report.
In May 1900, Gait sent a circular to district officers explaining that part of the information collected in 1891 was dubious because it had been impossible “to ascertain [some people's] true caste or tribe”. He supplied a list of the worst cases.Footnote 23 Gait received several hundred pages of letters responding to his circular, some confirming that the 1891 list was accurate or nearly so, whereas others – especially from districts in east Bengal, and Bihar and Orissa – reported errors and raised further questions. He circulated further lists in late 1900 and received more letters reporting mistakes. Gait's marginal comments show how seriously he took seemingly minor issues; thus, to cite one instance, he asked for clarification about whether ‘Doai’ and ‘Daoi’ were really distinct groups in Faridpur (east Bengal).Footnote 24
Some replies were pettily pedantic, but others raised real ethnographic questions. For example, sixteen letters mentioned the problematic case of Sudras. In Birbhum (west Bengal), the title ‘Sudra’ was not the name of a particular caste, but it was, with more or less certainty, in several districts in east Bengal. Thus in Dacca Sudras were ranked below Kayasthas and were a caste of servants to Brahmans, Baidyas and Kayasthas, whereas in Mymensingh they were descendants of “maidservants by their masters of good caste” and were also called Golam Kayasthas. Yet elsewhere in east Bengal, such as Noakhali, Sudra was not a true caste name; it was used for servants of the higher castes, but Sudras were “to all intents and purposes” Kayasthas, who probably invented the name for men “they did not like to associate with as equals”. In Orissa, Sudras were just a Chasa subcaste in Cuttack, whereas about half the Sudras in Puri were Chasis trying raise their status.Footnote 25
Gait showed the same concern with empirical accuracy when complying with Risley's request to investigate “social precedence”. Gait acknowledged that caste rank was fundamentally a matter of “Hindu public opinion”, though this was hard to determine because Hindus, he said, did not care about the status of castes other than their own and the only point of general agreement was that Brahmans are at the top.Footnote 26 Nonetheless, rank could be worked out by using five “well recognised tests of social position”: whether good or degraded Brahmans serve as priests; rules about taking food and water; whether touch is polluting; ceremonial observances (for example, whether widow remarriage is permitted); and availability of barbers’ and dhobis’ services.Footnote 27 In Bengal proper, the tests generated seven categories, from Brahmans at the top to untouchable “unclean feeders” at the bottom; the categories varied somewhat between regions.Footnote 28 Within each category, because their mutual ranking was normally disputed, Gait listed the main individual castes alphabetically; he put other small castes in a full table in an appendix. The accuracy of lists of ranked castes did not depend solely on the “social position” tests, however, because committees of “native gentlemen” were also consulted. Yet they did not always agree and trying to sort out social precedence often caused an “extraordinary amount of ill-feeling and jealousy”.Footnote 29
Gait's discussion of caste rank reflected the tension between social precedence as determined objectively by his five tests and its often variable, vague and contested character. A comparable tension existed between an ideally comprehensive list of castes and the acknowledged fact that some people told enumerators their subcastes, or other affiliations, rather than their ‘true’ castes. The nub of the issue at stake here can be clarified by looking at how Gait tackled the difficult problem of defining ‘caste’. In his 1901 Bengal report, Gait said it was hard to distinguish castes from the endogamous subcastes into which they are normally divided, partly because groups periodically split or merge, although “the real touchstone by which a decision is to be arrived at seems to be the general public opinion at the present time”.Footnote 30 He then set out this descriptive definition: “A caste is an endogamous group, or a collection of such groups, bearing a common name who, by reason of similarity of traditional occupation and reputed origin, are generally regarded, by those of their countrymen who are competent to give an opinion, as forming a single homogeneous community, the constituent parts of which are more nearly related to each other than they are to any other section of society”.Footnote 31
Gait also quoted Risley's descriptive definition in his Manual of Ethnography, which closely resembled that in the 1901 census report, which in turn was printed alongside a comparable one by Émile Senart.Footnote 32 Senart was a French Indologist, whose book on caste, which Risley admired, partly relied on the census reports. Although Senart was an evolutionist, he was a radically modern thinker in some respects, who insisted that the “past history of caste is only intelligible in the light of present conditions”.Footnote 33 Moreover, his description of an all-Indian caste system constituted by a series of features – principally endogamous marriage, traditional occupations, purity rules and ritual markers, community government, and hierarchy – was coherent and original in 1896. Possibly, Risley adapted Senart's text for his own definition of caste, which Gait revised again. Oddly, hierarchy was not actually part of any of the definitions, despite the attention all three writers gave it. The key difference between Gait's definition, and Risley's and Senart's, however, is that his included the phrase “generally regarded . . . to give an opinion”. Gait slightly modified his definition in an encyclopaedia article published in 1910 and once again in his 1911 census report, where he added the comment that “the decision as to what does, and what does not, constitute a caste is largely a matter of degree”.Footnote 34
Gait's explicit recognition of this problem distinguished him from Risley and other census officials in 1901 or earlier, except for Burn. Thus Gait and Burn were, I believe, the first official anthropologists to see that how people identify the castes or other social units to which they and others belong – as well as how they evaluate their own and other castes’ status – depend on subjective perceptions that are often variable and disputed. In other words, in the language of modern Weberian sociology, Gait and Burn recognised that the actors’ own intersubjective meanings are necessarily constitutive of institutions such as the caste system, which cannot be fully and objectively described as ‘social facts’ in Durkheim's sense.
Gait's approach to the evolution of caste should also be mentioned briefly. Gait broadly agreed with Risley's racial theory as provisionally supported by anthropometric data in 1901.Footnote 35 But he said little about origins in his encyclopaedia article or his 1911 census report, where he was also reluctant to comment on the “controversial question” of Risley's racial theory, although he was more openly sceptical in a lecture delivered in London.Footnote 36 Gait, it appears, was not much interested in caste origins, and nor were Blunt and O'Malley, although none of them ever actively rejected evolutionism and conjectural history.
As the 1911 census Commissioner, Gait oversaw a reduction in the amount of attention paid to caste compared with the three previous censuses. In his general report, Gait explained that the provincial reports contained “comparatively little fresh information” on caste, except for caste government and panchayats, which were a special topic for investigation in 1911 because more information on local government was wanted. Gait thought there was “still ample room for elaborate monographs . . . on the more important castes and tribes, but so far as a general description of them is concerned, comparatively little remains to be done”, given the copious data in previous census reports, as well as in the tribes and castes handbooks already published or in preparation.Footnote 37 Gait repeated this point in his London lecture, but also added that other kinds of ethnographic information were needed, especially about “the working and ramifications of the caste system and the dynamics of caste”.Footnote 38 These certainly were important questions, which Blunt and O'Malley in particular would later address. By 1911, Gait was the most experienced census official in India and his view that the censuses should change, because further information on caste was superfluous, had the weight of government behind it.
The 1921 and 1931 census reports included the standard tables of castes, but very little new material, except for data collected prior to the 1935 Act about Untouchables, which I discuss further below. In the 1931 census reports, Untouchables were called the “Exterior Castes”, instead of “Depressed Classes”, which was then the normal designation; soon, of course, they would be renamed ‘Scheduled Castes’.Footnote 39 At the 1941 census, the basic data on caste were collected, but not tabulated for publication in the general report, which instead contained one table for “community” classifying the population by religion. The Scheduled Castes appeared under “Hindus”; the table also included an extra category for hill and forest tribes, in principle corresponding to the Scheduled Tribes.Footnote 40 The last three colonial censuses actually collected less data than the earlier ones on most topics, but the reduction was particularly noticeable for caste, given its earlier prominence.
Religion and the 1911 Census
Before turning to Blunt's and O'Malley's work on caste, I must explain that religion was by far the most controversial matter at the 1911 census. After the Morley-Minto reforms, Muslim political leaders were concerned about their population numbers, but so, too, were their Hindu counterparts, who wanted to ensure their community was fully represented on the new legislative councils. The contentious question was whether ‘Hindus’ included Sikhs, Jains, Untouchables excluded from social intercourse by high-caste Hindus, Muslim or Christian ‘converts’ who were not truly Muslim or Christian, and tribal animists who were partially Hinduised. Prior to the census, Gait sent a circular to provincial superintendents about “misleading” census returns that “include millions of people who are not really Hindus at all’” which appeared in a Lahore newspaper in 1910 and provoked an uproar among Hindu leaders.Footnote 41 Gait alluded to the controversy in his census report, while explaining that superintendents were asked to report on how ‘Hindu’ could be defined in their own provinces.Footnote 42
In the United Provinces, agitation about religion affected the enumeration in 1911.Footnote 43 In his report, Blunt discussed the definition of ‘Hindu’ at some length, arguing that various more or less definitive criteria exist, notably recognition of Brahman supremacy and veneration of the cow. Even so, the definition was “essentially indefinite, and to define the indefinite is a contradiction in terms”.Footnote 44 Enumerators recorded as Hindus those who said they were, though often after prompting, but sharply dividing ‘Hindu’ from ‘non-Hindu’ was impossible, and ‘Hinduised’ fitted many tribal ‘animists’ better.Footnote 45 Because Islam was uniform in its ‘essential beliefs’, defining ‘Muslim’ was easier, but many Muslims held ‘unorthodox’ beliefs and many Muslim converts retained Hindu customs.Footnote 46 O'Malley made no mention of political agitation in Bengal, but did refer to problems about the enumeration of partially Hinduised tribal people.Footnote 47 He discussed the definition of ‘Hindu’ more briefly than Blunt, although he similarly identified Brahman supremacy and objection to cow slaughter as salient criteria; he also explained that “extraordinary divergence of opinion” about the criteria was found among Hindu spokesmen.Footnote 48 O'Malley's observations about Muslims resembled Blunt's.Footnote 49
The diverse views of Blunt, O'Malley and other superintendents informed Gait's discussion about how to define ‘Hindu’; he also pointed out that “difficulties of classification” were not confined to Hindus, and mentioned various communities whose mixed Hindu and Muslim traits exemplified “the extremely indefinite character of the boundary line between different religions in India”.Footnote 50 Most students of popular religion in India would agree that Gait's conclusion was and still is correct. It also reiterated what many other census officials, in addition to Blunt and O'Malley, wrote in 1911 or earlier, notwithstanding the tables of supposedly exact religious population figures. Yet this was a statement by a senior British official throwing serious doubt on the claim that religious difference separated Hindus and Muslims into distinct communities. In the wake of the Morley-Minto reforms, Gait must have understood his statement's implications, just like Athelstane Baines (the 1891 census Commissioner), who commented that Gait's lecture about the 1911 census “had shown most lucidly that no uniformity should be inferred from this [religious] community of designation”.Footnote 51
It is true that a range of social and cultural features distinguished Hindus and Muslims more or less sharply and they could look back on different political histories, especially in the old Mughal heartland of northern India. Moreover, Muslims almost always identified themselves as Muslims, whereas Hindus did not always self-identify so unequivocally. The British, however, did not just emphasise religious difference; their “historical game-changer”, to quote Peter Gottschalk, was to describe the two religions as “mutually antagonistic communities” that were also “mutually exclusive categories of social belonging that bifurcated nearly every societal and cultural dimension of India”.Footnote 52 This comprehensive discourse underpinned the separate religious representation brought in by the legislative reforms, which progressively created new political realities, communities and categories inconsistent with the more heterogeneous social ones that the censuses and official anthropology recorded.
Blunt on Caste
In his 1911 United Provinces census report, Blunt wrote a long chapter on caste, which – following Gait's instructions – included a detailed ethnographic account of caste government and panchayats.Footnote 53 It also notably included the first, albeit brief, description of jajmani, the traditional village economic system based on enduring patron-client relationships.Footnote 54 When on leave from his administrative duties, Blunt wrote The Caste System of Northern India, which was published in 1931. Quite a lot of this book derived from his census report, but Blunt also modestly claimed he had merely rearranged William Crooke's tribes and castes handbook, which provided copious information about every group, but never “a full and connected account of caste as a system”.Footnote 55
Blunt, who began by explaining that heredity, endogamy and commensal restrictions were the defining characteristics of caste, endorsed Gait's 1910 encyclopaedia definition, while slightly expanding it.Footnote 56 After conventionally outlining the evolution of caste, Blunt turned to the ethnography. His detailed examination of caste, subcaste and marriage broke new ground, but he also discussed commensal restrictions, caste government, and the occupational and economic aspects of caste.Footnote 57 Other chapters covered different types of caste, and caste and Islam.Footnote 58 In a generally pedestrian treatment of “caste in relation to religion”, Blunt stressed the Brahman's social and religious primacy. He examined conflicts between caste custom and written law more perceptively, and ended with modern change.Footnote 59
Blunt's book was not very well organised, presumably because it was written at intervals while he was a serving ICS officer, though that gives some of it an engaged immediacy. Nor was Blunt always persuasive; for instance, he underplayed the importance of status and hierarchy, partly in reaction to the furore over ranking at the 1901 census and partly because he was so preoccupied with caste and occupation. Nonetheless, the heart of his book was an examination of the structure and workings of the caste system in a specific locality, the United Provinces, at a specific time, the early twentieth century. Blunt was rather hazy about how systematic the caste system actually was, but more than any previous study, his text was in practice a predominantly functionalist analysis based on ethnographic evidence.
In one short but important passage, Blunt highlighted “the segmentation of Hindu society” and saw, as Gait did not, that it explained “the ignorance of the average Hindu about his caste system” and the tendency to give different answers when asked to name his caste. Thus, for example, someone might answer with his varna (the classical divisions of Brahman, Kshatriya, Vaishya or Shudra), his ‘caste’ or ‘endogamous subcaste’, his ‘exogamous section’, or his ‘caste-title’.Footnote 60 These variable referents of indigenous terms such as jati were similarly identified by André Béteille when first explaining that “people view themselves as belonging to units of different orders in different contexts”, so that the caste system must be understood as a segmentary system, not misconstrued as an array of separate, reified groups.Footnote 61 Louis Dumont also emphasised the caste system's segmentary nature, although he somewhat confused matters by reopening the old debate about whether the caste or subcaste is the ‘real’ sociological unit, citing Blunt among others.Footnote 62 Dumont, in fact, cited Blunt frequently, but on segmentary systems he and Béteille both invoked E.E. Evans-Pritchard's classic study of an African tribe, not Blunt's book.Footnote 63 It was a telling example of how post-Independence scholars consistently found more inspiration in modern anthropologists of other parts of the world than in their colonial predecessors.
O'Malley on Caste
Like Blunt, O'Malley wrote a long chapter on caste in his 1911 census report on Bengal, which included a detailed historical and ethnographic examination of caste government throughout the province.Footnote 64 After retiring from the ICS in 1924, he wrote several books in England in the 1930s about Indian society and history. O'Malley normally wrote with academic detachment from his subject matter and was exceptionally well read in the colonial anthropological literature. He began Indian Caste Customs by outlining the caste system's “salient features”, which include hierarchical ranking, commensal restrictions, hereditary occupations, and most importantly endogamy. He explained how castes are divided into subcastes and exogamous kin groups, and emphasised that caste is religious as well as social, for it is “the steel frame binding together the many beliefs massed together in Hinduism”.Footnote 65 In his book on Hinduism, O'Malley similarly stressed the connection with religion, describing the caste system as “a powerful moral force” and “Hinduism in its social form”.Footnote 66 In discussing caste, O'Malley gave relatively more weight to religion than other official anthropologists and, unlike Blunt, he followed Risley in emphasising hierarchy. The thematic chapters of his book successively examined caste government, marriage and commensality rules, occupations and, finally, the social condition of Untouchables and modern change.Footnote 67
Another book, India's Social Heritage, not only discussed caste, but also tribal groups, as well as the village community and the family.Footnote 68 In his chapter on Hindu society in Modern India and the West, an edited volume published posthumously, O'Malley identified the chief Indian social institutions as the family, caste and village community, and examined how they had changed under British rule.Footnote 69 Interestingly, David Mandelbaum's major work also took family, caste and village to be the three “principal institutions” of Indian society, but whether Mandelbaum was directly influenced by O'Malley is unclear from his introductory acknowledgements.Footnote 70
Indian Caste Customs is concise and contains less ethnographic detail than Blunt's book, but is better written. O'Malley was the first official anthropologist to state clearly that the caste system is indeed pan-Indian and he may have been inspired by Senart's “brilliant monograph”.Footnote 71 Moreover, he argued, the system is predicated on a relational logic; throughout India, despite regional diversity, “caste remains the basis of the social order, with its numerous divisions, each of which has a social value in relation only to other divisions . . . The differences are of form and not of substance [and] there is a fundamental unity of system”.Footnote 72 This passage led Dumont to compare O'Malley with Evans-Pritchard, although O'Malley never specifically referred to segmentation.Footnote 73 Most importantly, O'Malley was the first author to write about the caste system in something akin to a modern, structural approach, as well as the first to do so without paying any attention to its origins and evolution. Hence, much like Blunt's book was in practice functionalist, O'Malley's was a predominantly synchronic, proto-structuralist exposition of caste as an integrated system.
Whether Blunt and O'Malley knew anything about contemporary anthropology in Britain and its emerging structural functionalism is unclear. Certainly, though, they were the two colonial anthropologists whose writings most fully anticipated later studies of caste and were sometimes cited as valuable sources, for example, by Dumont.Footnote 74 On the other hand, the social scientists investigating local, village caste systems after Independence mostly distanced themselves from earlier work, partly because it was seen as colonial and old-fashioned, and partly because Indian anthropology in the inter-war years, as pursued by both British officials and a new generation of Indian academic researchers, was primarily the study of ‘primitive’ tribes. It is probably salient, too, that the final work on caste by an official anthropologist was J.H. Hutton's Caste in India in 1946.Footnote 75 Hutton, the 1931 census Commissioner and doyen of Indian tribal anthropology, was the professor of social anthropology at Cambridge University from 1937 to 1950, so that his book carried the imprimatur of academic authority. But much of it was actually about tribes, not castes, or consisted of evolutionist-cum-diffusionist conjectural history, which modern, functionalist anthropologists had condemned by the 1940s. Hutton, therefore, never inspired the post-Independence generation and may even have pushed the work of earlier ICS scholars, including Gait, Blunt and O'Malley, into the shade.
Gait, Blunt and the Montagu-Chelmsford Reforms
In the last part of this article, I look at the relationship between colonial anthropology and political developments from 1911 to 1939, in which Gait and Blunt (though not O'Malley) were active figures.
In 1912, Gait was transferred to Bihar and Orissa, the new province created when the partition of Bengal was reversed, and soon became a member of the Lieutenant-Governor's executive council. From 1915 until his retirement in 1920, Sir Edward Gait, as he now was, was himself the Lieutenant-Governor. In this capacity, he took part in extensive consultations between 1916 and 1919 about the Montagu-Chelmsford proposals for constitutional reform, which were intended to promote “responsible government in India as an integral part of the British Empire”.Footnote 76
Like almost all the British ruling elite, Gait was contemptuous of Indian educated, urban, middle-class politicians who allegedly represented hardly anybody but themselves, and certainly not the rural ‘masses’. In 1916, Gait cautiously favoured reform, but opposed any changes, such as abolition of separate electorates for Muslims and landlords, that could enhance the power of “lawyer-politicians”.Footnote 77 He later supported more substantial change, because he and his council in Bihar and Orissa became convinced that the Great War had turned into “an all-world struggle between the principles of democracy and autocracy”, so that “the greatest possible liberalisation of British rule in India” was required.Footnote 78 In November 1918, on the eve of Armistice Day, another letter from Gait and his council insisted that the war had strengthened the case for self-determination and expressed the view that British rule could not last indefinitely, because “it is contrary to human nature that three hundred million people should acquiesce in the perpetual domination of a small body of foreigners”.Footnote 79 In January 1919, at a final meeting about the reforms, a split emerged between two heads of government – Gait and Lord Ronaldshay, Governor of Bengal – who supported the partial transfer of powers to provincial governments under the dyarchy scheme, and five others, who opposed it and were decried as reactionary “satraps” by Edwin Montagu, the Secretary of State, and Lord Chelmsford, the Viceroy.Footnote 80 By the standards of his day, and also unlike most members of the ICS, Gait had a relatively progressive outlook on Indian affairs, though he also shared the now prevalent assumption that separate Muslim electorates were necessary.
Throughout much of northern India at this time, the political division between Hindus and Muslims was deepening. Gait's own worst crisis arose in October 1917, when Hindus perpetrated massive violence on Muslims in the Bakr-Id riots in Shahabad District.Footnote 81 These riots, Gait told the Viceroy, were “wholly unexpected”;Footnote 82 his government, it seemed, had little cognisance of how religious communalism was developing on the ground.
In December 1917, W. S. Marris, a Home Department official, sent a circular to heads of provincial governments about the reforms, which included a note opposing separate communal electorates for the new legislative councils. Marris had chaired a United Provinces government committee on the reforms on which Blunt also sat. Blunt wrote one paper about the “classes of the people” describing the political aspirations of the “landed aristocracy”, “lesser landed gentry” and urban educated professionals, and their absence among the rest of the people; he did not refer to caste divisions at all. He wrote a second paper strongly criticising communal representation, which formed the basis of Marris's note. Blunt argued that other caste and religious groups besides Muslims – such as non-Brahmans in Madras or Arya Samajis – would inevitably demand communal representation as well. Refusing their demands would be illogical, but the outcome was bound to be disastrous, because disputes between religious communities and castes would intensify, making civic spirit and responsible government into hopeless prospects. Creating Muslim electorates in 1909 was “a serious mistake” and, even if correcting it were unrealistic, retaining communal representation could not be recommended; he concluded: “Let the responsibility for healing the feud rest on the antagonists”. The “feud”, Blunt obviously thought, could not be blamed on the British, but they should not make it worse. Marris's note, however, also suggested a compromise form of proportional representation for Muslims.Footnote 83
Gait was well aware of the potential dangers of caste-based politics in Bihar.Footnote 84 Nevertheless, he and his council flatly opposed Marris's (and hence Blunt's) arguments about communal electorates, insisting that Muslims could be properly represented in councils only through separate constituencies, because even though they live alongside Hindus, they “are a separate community with entirely different religious beliefs and social customs”, whose interests often clash with those of Hindus.Footnote 85 Gait's expression paraphrased what Risley had written in 1908 to justify separate Muslim electorates; the Muslims, declared Risley, were “an absolutely separate community” from Hindus by social and even racial criteria, as well as religious ones.Footnote 86 Yet Risley must have known the ethnographic evidence did not support what he wrote.Footnote 87 The same must have been true of Gait when, in 1917, he contradicted his own conclusions in the 1911 census report about the indefinite boundary between religions. There are, I think, two likely explanations for Gait's apparent bad faith. The first is that Gait knew his letter to Marris was factually incorrect, but, as a high government official, he deliberately exaggerated the differences between Hindus and Muslims, partly for reasons of Realpolitik and partly because he assumed no further reforms could happen unless separate Muslim representation was preserved. The second is that he decided his conclusions in 1911 no longer held good in Bihar after the Shahabad riots, because separation between Hindus and Muslims, and their respective religious customs, was sharper than it had been or he had previously realised. I do not know which explanation is more likely, but Gait's apparent inconsistency highlighted the gap between political realities and colonial anthropological knowledge which was growing throughout this period.
Blunt and Politics in the 1920s and 1930s
After the war, Blunt held a series of posts in the United Provinces government and served on the Governor's executive council from 1931 to 1935; duly knighted, he retired from India in 1936. On the whole, judging by his history of the ICS, Blunt was well-disposed towards the Montagu-Chelmsford reforms and those brought in by the 1935 Government of India Act. He shared the paternalistic attitude towards the peasantry characteristic of civilians and deplored the district officers’ loss of authority under dyarchy, so that “the welfare of the people for which [they] were still responsible” was sacrificed “to political expediency”.Footnote 88 O'Malley, in his history of the ICS, voiced the same criticism, but more vehemently, and he was plainly more hostile to the reforms than Blunt.Footnote 89 Despite his complaints, Blunt thought the ICS had successfully managed the reformed system in the twenties and thirties, and he worked extensively with Indian non-officials, many of them elected politicians, on the legislative councils set up by dyarchy. He described relationships between British officials and moderate politicians as generally cordial, even though, from the end of the war until 1935, “bitter animosity” prevailed among “extremist politicians” and “constantly recurring irritation” among the moderates.Footnote 90
The 1935 Act brought in special measures to secure political representation for Untouchables, and this was probably the most critical issue on which Blunt and other official anthropologists provided data and advice to the government in the inter-war period. Mainly owing to the work of Christian missionaries and Hindu social reformers, the Untouchables’ deplorable conditions of existence were a matter of serious public concern by the beginning of the twentieth century. At this time, the English terms ‘Untouchable’ and ‘Depressed Classes’ first became current to denote a pan-Indian category of people previously known by local caste names. For Hindu political leaders, counting Untouchables as Hindus became important after the Morley-Minto reforms, as we have seen. But who the ‘Untouchables’ were remained unclear, especially in north India, where the practice of untouchability was less severe than in the south, and one consequence was divergent official estimates of their population.Footnote 91
By 1930, the Depressed Classes were firmly on the political agenda and the report of the Indian Statutory (Simon) Commission, which reviewed the workings of the Montagu-Chelmsford reforms, specifically discussed their “disabilities” and “pitiable” state.Footnote 92 B. R. Ambedkar, the Untouchables’ leading spokesman, called for their political recognition as a distinct, deprived minority, and powerfully presented their case to the Commission.Footnote 93 In 1932, as Ambedkar wanted, the Communal Award announced by the British government granted separate electorates for the Depressed Classes, alongside Muslims and other minority groups. But Gandhi protested with a ‘fast to the death’ and Ambedkar, in the Poona Pact, then agreed to give up separate electorates for a larger number of reserved seats within the general, Hindu constituency; this was the system eventually included in the 1935 Act.
The precise identity of the Depressed Classes remained in question, however, and had to be resolved before the reformed constitution could be implemented. Helping to solve this problem had been a task for the 1931 census officials. Before the census started, Hutton told provincial superintendents they would have to produce lists of the Depressed Classes, which would undoubtedly be difficult, but it was agreed that “each province should make a list of castes who suffered disability on account of their low social position and on account of being debarred from temples, schools or wells”. Furthermore: “No specific definition of depressed classes was framed and . . . because it was realised that conditions varied so much from province to province and from district to district’, superintendents were not given ‘meticulous instructions”.Footnote 94 In his general census report, Hutton discussed the definition of “Exterior Castes” – his preferred term, originally coined by C. S. Mullan, the Assam Superintendent – and acknowledged that a wide range of tests of untouchability could be applied, but from the state's viewpoint, “the important test is the right to use public conveniences – roads, wells and schools”, so that “religious disabilities” and “social difficulties” were contributory factors only.Footnote 95 In the end, lists of Exterior Castes, with their populations, were produced for each province, and an all-India total was calculated, so that the government's requirements were met. All the same, superintendents interpreted their instructions variously, reached different conclusions about untouchability and its definition, and used alternative tests deemed suitable for their own provinces, which were not necessarily Hutton's preferred ones; their ethnographic material, more copious in some reports than others, also clearly showed that the custom and practice of untouchability was highly diverse.
The Indian Franchise (Lothian) Committee, whose members included Ambedkar, effectively discounted much of what Hutton wrote, without explicitly saying so. It selected just two “generally accepted tests of untouchability”: denial of access into Hindu temples, and causing pollution by touch or within a certain distance. Importantly, too, the Committee unambiguously equated the “Depressed Classes” with castes that were untouchable by these two tests.Footnote 96 Both were extracted from Gait's ten criteria, listed in the religion chapter of the 1911 census report, for identifying “partially assimilated Hindus”, whose enumeration as “genuine Hindus” was problematic. The tests were not meant to define untouchability or to classify castes, although the Franchise Committee report – presumably relying on Ambedkar's opinion – implied that they were.Footnote 97
One memorandum submitted to the Franchise Committee was Blunt's note on the Depressed Classes, which examined the problem of their definition in the United Provinces.Footnote 98 Blunt contended that “untouchable” and “depressed” should not be regarded as synonyms. He also explained that the criteria for determining untouchability were not straightforward: for example, the condition could be permanent or temporary; many castes were untouchable to Brahmans, but not other castes; Untouchables often stigmatised Untouchables from other castes; and social conditions were changing rapidly, so that untouchability was being mitigated in many circumstances. There are, Blunt concluded, a few castes that were always untouchable, at least to all Brahmans, and their definite untouchability could reasonably be regarded as making these castes “depressed”. In practice, though, the criteria for defining “depressed” have been “so wholly vague and inconclusive” that anyone could give it almost any meaning they wanted. In the final analysis, the criteria were really political: the Depressed Classes, said Blunt, were those which could not secure adequate political representation and protection of their interests without special franchise concessions, and they were not all Untouchable castes. Furthermore, they had to be distinguished from the equally ill-defined “Backward Classes” made up of “lower Sudra groups”, which could secure political representation and protection through the general electorate, but were unlikely to do so. Blunt ended his note with a list of castes in the United Provinces, indicating which were “untouchable”, and classifying all of them as “depressed”, “backward” or neither.
Shortly after Blunt, Ambedkar wrote two notes for the Franchise Committee; only the second, shorter one was published.Footnote 99 In discussing the United Provinces, Ambedkar responded at some length to Blunt, whose expertise he acknowledged. In his first note, Ambedkar particularly emphasised the tripartite distinction between “primitive tribes” outside Hindu religion and society; “backward”, “lower-class Hindus” within them; and Untouchables within Hindu religion but outside Hindu society. This last group was appropriately called “Exterior Castes”, but generally designated the “Depressed Classes”, a term that should be taken as synonymous with “Untouchables”. In both notes, Ambedkar argued that despite great diversity across India, it would be wrong “to suppose that differences in tests of untouchability indicate differences in the conditions of the untouchables”, for all of them alike “suffer from social odium”. He also insisted that modern change had brought about hardly any real improvement. In his second note, he developed that point by drawing a distinction between Untouchables in the “literal” sense and those in the “notional” sense, who belonged to a caste deemed to be polluting even if their touch did not actually pollute in all contexts. Yet the distinction was ultimately immaterial, for “untouchability in its notional sense persists even where untouchability in its literal sense has ceased to obtain”, mainly because it was rooted in Hindu religious dogmas. For all these reasons, Ambedkar rejected both Blunt's restricted definition of untouchability, and his distinction between the Untouchables and Depressed Classes.
In the end, of course, the Depressed Classes were equated with the Untouchables and Ambedkar's conclusions were largely accepted, despite disagreements within the Committee about the identification and population size of Untouchable castes in northern and eastern India (which Ambedkar's first note discussed at length). In 1936, the Depressed Classes became the Scheduled Castes and were listed to give effect to their special electoral representation in the post-1935 constitution. In practice, “the list reflected definitions of untouchability with an admixture of economic and educational tests and considerations of local politics”, as Marc Galanter explained, but it was, and in its later versions remains, a list axiomatically defined by officially endorsed caste status.Footnote 100
Although the Franchise Committee acknowledged Hutton's and Blunt's expertise, it did not in the end take much notice of the information and advice received from the official anthropologists. The 1931 census certainly proved that Untouchables suffered from appalling privation and discrimination in much of India. But it also showed that untouchability varied greatly from place to place, no pan-Indian group of Untouchables sharing a common social position actually existed, and – unless the phrase were defined very loosely – there was no universal “social odium”. The politics of untouchability, caste and religion were all shaped by a reified, classificatory logic partly created by official anthropology, but the ethnographic evidence also revealed the logic's mismatch with diverse social reality. Blunt's anthropological analysis of untouchability in the United Provinces made good sense. Nonetheless, Blunt himself recognised that the question of the Depressed Classes was fundamentally political, and he probably never could have persuaded Ambedkar and the Franchise Committee to distinguish between Untouchables and Depressed Classes, or to be more precise in defining them.
It is interesting that Blunt, in concluding his book on caste in 1931, disputed the assumption made by almost all other colonial anthropologists, including O'Malley, that in India's fragmented society “caste and nationality are . . . incompatible”, arguing that “through the diversity of Hindu society” the one bond uniting its many parts had been “the Brahman hegemony”. Blunt, as we saw, regarded the Brahmans’ acknowledged supremacy as a crucial feature of caste and Hinduism. National and caste interests sometimes clashed in India, he continued, but not very dangerously, because “the leaders of society will also be the national leaders”. Much would therefore depend on the Brahmans, since “every single Hindu political leader of first class importance in this province is a Brahman”.Footnote 101 I do not know why Blunt took such an elitist line in his book, so that he ignored the different interests of the lower castes, as well as Muslims, and why, particularly given his antipathy to the extremists, he was so impressed by Brahman politicians. But by 1932, he seems to have partly changed his mind. Blunt described the Untouchables’ plight as worse in his note than his book.Footnote 102 More significantly, he commented that upper-caste Hindu leaders wanted the support of the Depressed Classes, but may “have allowed the propitious moment to slip by unheeded” because the latter were developing their own political consciousness: “The break between the two [caste groupings] is not yet complete, but it has begun and is continuing”.Footnote 103
In relation to Hindus and Muslims, it is hard to tell from Blunt's writings how much of the pessimism that he expressed about India's political future during the Montagu-Chelmsford debate persisted into the thirties. After the war, Blunt probably witnessed communal tension in the United Provinces for himself and he must have known a lot about it; in 1931, for example, just before he finished his book on caste, serious Hindu-Muslim riots erupted in Kanpur. In his study of the ICS, he described the “age-old antagonism between Hindus and Muhammadans” as “an ever-present source of anxiety to the district officer”, although he claimed that rioting was not normally very difficult to control.Footnote 104 After quoting the parliamentary committee report on the 1935 legislation, which highlighted Hindu-Muslim antagonism and rigid caste divisions incompatible with democracy, Blunt also described the post-1935 constitution as “beset with difficulties”.Footnote 105
But Blunt left it at that and said little more in his last publication, an edited volume that provoked a controversy germane to my discussion. In 1936, a committee on the training of ICS recruits during their probationary year recommended introducing a compulsory course of lectures on “social welfare”.Footnote 106 Despite earlier initiatives, the study of Indian society or of anthropology had never previously been a required part of ICS training. Blunt (who was now retired) chaired a second committee, which devised the syllabus and agreed to produce a textbook for the course.Footnote 107 The India Office then decided to publish a revised version for sale to the general public, and Social Service in India, with two chapters on geography and social structure by Blunt, eventually appeared in early 1939.Footnote 108 But the book, specifically the discussion of caste and Hinduism in Blunt's chapter on social structure, immediately upset A. Ramaswami Mudaliar, a former non-Brahman politician in Madras, who was a member of Blunt's committee and of the Council of the Secretary of State for India. Mudaliar, who itemised the objectionable paragraphs in a letter to the Under-Secretary, said the book would be “sharply criticised in India”, where people were “oversensitive” about “caste, religion and social life”, and it could provoke an outcry as big as Katherine Mayo's Mother India, which had outraged Indians twelve years earlier. He worried, too, that because Blunt wrote much more about Hindus than Muslims, the discrepancy could be deliberately misinterpreted, with “disastrous” consequences in a time “surcharged with communal tension”. Blunt and the Under-Secretary quickly decided to withdraw the book, so that it could be altered.Footnote 109 But instructions to retrieve the copies sent to India were not followed properly, which angered officials in London, because some Indian editors “may now proceed to make capital” out of the affair, and a couple of hostile newspaper reviews did appear, particularly criticising Blunt.Footnote 110
Blunt worked rapidly to change everything that Mudaliar had criticised, except for the imbalance in attention given to Hindus and Muslims. The modified version of the book was then published. Viewed against the political backdrop of 1939, the ructions over Social Service in India were trivial. Yet it is striking how quickly and fully Mudaliar's objections were met. Blunt and other officials immediately saw the political dangers of publishing material on caste and religion that might offend “oversensitive” Indians. The affair also showed clearly that colonial anthropological knowledge about traditional caste, religion and the family – however accurate or scholarly it was – had become thoroughly detached from political realities that no government publication could ignore by 1939.
Blunt must have been aware of this, despite the misjudgements he had made, because his discussion of modern change was notably terse about caste politics – mentioning only protests against recording caste at the 1931 census – and virtually silent about Hindu and Muslim communalism; Mudaliar did not criticise these lacunae. In his last optimistic paragraph about an “Indian nation . . . now in process of formation”, in which caste would probably survive, but was beginning to die, Blunt merely stated that “old racial differences” between Indians and Europeans “have been replaced by communal differences”.Footnote 111 Leaving out politicised caste and religious communalism was undoubtedly judicious. It is telling, though, that a book about “some social and economic problems” originally designed for ICS trainees, both British and Indian, avoided these deepening political problems. The omissions also revealed that Blunt – like Gait earlier and before them Risley – was uncertain or even plainly inconsistent when discussing political issues in the light of anthropological knowledge.
Conclusion
In 1876, when they were still junior officials in India, Ibbetson and Risley submitted evidence to the official Salisbury committee on ICS training.Footnote 112 They both stressed the importance of reading Henry Maine, and Ibbetson also mentioned John Lubbock, J.F. McLennan and Tylor, all leading anthropologists in Britain, whose books were indispensable for what he called “archi-sociology” (a term that never caught on). Ibbetson and Risley were convinced that civilians should possess an intellectual understanding of India, not just bureaucratic competence; they also believed that ethnographic data collected in India could vitally contribute to modern anthropology (or ethnology or sociology), as well as to good government and administration. This double purpose, academic and political, was forcefully expressed by Ibbetson in his 1881 Punjab census report, where he explained that he had expressed his own views on caste and religion in particular, because he hoped to draw attention to the “extraordinary interest” of the abundant material that could be “of such immense value to students of sociology”. Moreover, he added, British “ignorance of the customs and beliefs of the people among whom we dwell” not only “deprive[s] European science of material which it greatly needs, but it also involves a distinct loss of administrative power to ourselves”.Footnote 113
Ibbetson and Risley, as their writings show, were excited about anthropology in India. They were also confident about the rightness of the British Raj, although the convictions of Ibbetson, who ended his career as the Lieutenant-Governor of the Punjab trying to repress nationalist ‘sedition’ in 1907, eventually become reactionary stubbornness, while Risley, who had drafted the key documents for the Morley-Minto reforms and retired in 1910, showed himself to be more pragmatically flexible. After the watershed of the war and the Montagu-Chelmsford reforms, Victorian imperialist attitudes persisted among many older civilians, but many others – like Gait and Blunt – accepted the need for political reform. By 1935, the declared goal of British rule in India was dominion status leading to self-government, which Blunt supported, but he assumed, like most of his countrymen, that the British would remain for a long time. Thus the ICS would still need British recruits, who would have to be trained to equip them for the new political dispensation; as Blunt explained: “The civilian who used to serve by ruling, must learn to rule by serving”.Footnote 114 Finally, six decades after the Salisbury committee, the Blunt committee ensured that young ICS officers would have to be taught about Indian society and probably learn some anthropology. But the reasons were very different from those put forward by Ibbetson and Risley, and expressed in Ibbetson's ringing declaration in 1881 - when civilians certainly were the rulers - that without anthropological knowledge British colonial power could be lost. In any case, though, Blunt's book arrived too late; owing to the war, ICS training in England was suspended and British recruitment ceased, so that very few probationers ever had to read Social Service in India.
In discussing the impact of political change on the censuses and the formation of colonial knowledge, Gottschalk explains that when the last census was held in 1941, British officials knew full well that religion and caste had become highly politicised. Hence when enumerators asked people to identify themselves, they did so in a setting in which census officials “no longer controlled the process of selecting those identities because the categories of belonging had become so politicised by an increasingly involved public”. But the situation had already started to alter following the Morley-Minto reforms and the result was that the government's interest in reporting religion and caste statistics “waned as popular determination to be recorded rose”.Footnote 115 Gottschalk's observation about officials losing control of the census categories can be extended to the anthropology of religion and caste more generally, as the controversy over Social Service in India illustrates well. Gait, Blunt and O'Malley in particular, building on the work of Ibbetson, Risley and others, contributed significantly to the study of caste, though less so religion, in the early twentieth century. At the same time, though, their work, irrespective of its enduring scholarly quality and their own intentions, became increasingly separated from the political realities confronting the government of India and its civil servants. As the Raj declined, colonial anthropological knowledge could no longer reinforce the colonial state.
