The Case of Hindu-Muslim Conflict in India

Precolonial Ethnic Violence Abstract: Political scientists studying the origins of ethnic conflict in the nonwestern world often focus on European colonialism. However, few studies systematically engage with precolonial history. This article examines the influential case of Hindu-Muslim violence in India, often portrayed as the outcome of British “divide-and-rule” policies. We construct a new dataset of all conflicts between Hindu and Muslim polities from 1000–1850 A.D. and use it to generate an estimate of conflict stock. We find that historical violence correlates with contemporary riots beginning around 1700 A.D. – before the British ruled the subcontinent. We suggest that this statistical relationship is driven by ethnic counter-mobilization: the decline of the Mughal Empire and rise of the Marathas in the late 17th century sowed the seeds of Hindu nationalism, the proximate cause of contemporary riots. Our findings suggest that scholars of ethnic violence, especially religious conflict, have exaggerated the impact of colonialism. Word count: 10,652 words Keywords: ethnic violence; colonialism; historical legacies; Hindu-Muslim conflict; Hindu nationalism; India 1 On the whole except perhaps among Pakistani scholars, there has been a tendency to see the Hindu-Muslim conflict as a rather recent phenomenon. Is this not a very short-sighted answer? – Marc Gaborieau1 I Introduction A casual glance at ethnic violence in the world today brings to mind William Faulkner’s quip that the “past is never dead. It’s not even past.” 2 In the Levant, Jews and Arabs continue a lengthy battle over the holy land. Tribal conflict remains a problem across the African continent. Even after the end of a protracted civil war, Sri Lanka suffers ongoing tensions between the Sinhalese and Tamil communities. And in India, the rise of Hindu nationalism since the 1980s has led to waves of violence against the minority Muslim community. Political scientists studying the origins of these ethnic conflicts across the nonwestern world have often focused on the impact of European colonialism, a perspective that can broadly be summarized in three arguments. First, scholars contend that ethnic groups were ill-defined before colonial rule, and violence was rare in precolonial times because fluid boundaries between diverse communities promoted tolerance and coexistence.3 Second, colonial administrators then “constructed” modern ethnic groups through the state-making processes of conducting censuses, drawing maps, and building museums.4 Finally, colonial empires instituted policies of ethnic favoritism and “divide-and-rule” that ultimately gave rise to group conflict.5 Matthew Lange summarizes a broad literature in writing that for “the majority of Africa and Asia, the social processes constituting modernity only became a major transformative force in the late nineteenth and early twentieth centuries,” and therefore ethnic violence should be considered a “modern menace.” 6 What is noteworthy, however, is how little research on ethnic violence ventures back beyond the colonial era. Colonialism is taken to be the crit1 2 3 4 5 6 Gaborieau (1985: 8) Faulkner (1951) On the Indian case, see Roy (1984); Hasan and Roy (2005). Anderson (1983); Hobsbawm and Ranger (1983). Pandey (1990); Mamdani (2002) Lange (2017): 44 and 60 2 ical juncture for understanding conflict among ethnic groups, but political scientists rarely ask whether such conflict was already widespread during the reign of precolonial polities such as Vijayanagara, Siam, or the Kingdom of Mutapa.7 This reticence is odd, considering that social scientists, especially economists, have explored how deep historical legacies may explain other contemporary outcomes like democratization and development.8 James Fearon and David Laitin note about the study of ethnic violence, however: “ ...when political scientists have addressed the question of ‘deep historical roots,’ it has mainly been to debunk claims of ‘ancient hatreds’ in specific ethnic conflicts.” 9 Is precolonial history important in the production of contemporary ethnic conflict? And if so, why should historical myths, symbols, and images continue to have – often thousands of years later – instrumental power for elites, as well as motivational resonance with modern populations? Answers to these important questions require expanding our historical view beyond the narrow confines of the 19th century. This paper explores the topic of historical ethnic violence using the influential case of Hindu-Muslim conflict in India. Communal violence was at the center of India’s bloody partition, and riots claimed more than 7,000 (mostly Muslim) lives from 1950-1995.10 A variety of scholars – political scientists but also historians – have argued that this recurrent bloodshed originated during colonialism, especially due to British policies such as the introduction of a census in 1871, the Partition of Bengal in 1905, and the creation of separate Hindu and Muslim electorates in 1909. We argue, in contrast, that contemporary Hindu-Muslim conflict has precolonial roots.11 We theorize that Hindu and Muslim identities were constructed in precolonial India, and we draw especially on the work of scholars of Indian religions in making this case. The origins of Hindu-Muslim conflict date specifically to the late 17th century. During most of the Mughal period (1526-1757), 7 Foa (2016) For example, see Acemoglu and Robinson (2009) on democratization and Spolaore and Wacziarg (2013) on development. 9 Fearon and Laitin (2015: 6) 10 We use communal in this paper as synonymous with religious. Data on riots from Varshney and Wilkinson (2006). 11 This is a minority position in existing literature, but for scholars who have made this argument in various forms, see Bayly (1985); Gaborieau (1985); Mahmood (1993); Talbot (1995); Subrahmanyam (1996); Lorenzen (1999); and Kruijtzer (2009). We build on these works by studying the roots of Hindu-Muslim conflict through an econometric approach. 8 3 emperors embraced ecumenical policies that encouraged cooperation between the Muslim gentry and local Hindu rulers. Yet as the Mughal Empire reached its zenith, the emperor Aurangzeb (reigned 1658-1707) broke with this longstanding tradition and instituted a series of repressive policies toward Hindus, most notably the reintroduction of the jizyah in 1679, a tax on non-Muslims. The Marathas, a warrior clan in western India led by Shivaji Bhonsle (r. 1674-80), rebelled against the Mughals, and in doing so, explicitly used the Hindu religion, as well as Hindu symbols, imagery, and mythology, as a force to consolidate their uprising.12 We test our argument using a new historical dataset of Hindu-Muslim conflict. Our underlying source is the 26-volume British encyclopedia, The Imperial Gazetteer of India (1909 version; hereafter IGI). The IGI was written by British civil service officers, some of them well-versed in languages like Sanskrit or Persian, who created entries for each of India’s imperial districts. Critically, these entries often contain historical information on group conflict. We used a detailed coding protocol to compile IGI data on historical violence between Hindu and Muslim polities in each district. We employed multiple research assistants in order to test for intercoder reliability, and different iterations of our dataset are broadly correlated with one another. The novel feature of our dataset lies in its longitudinal nature. In contrast to other studies of the long-term persistence of violence that we discuss later, our focus is on conflict over the long durée. Our dataset covers the entirety of Hindu-Muslim interaction in precolonial India – beginning from the initial raids of Muslim conquerors like Mahmud of Ghazni and Muhammad Ghori in the 11th century, to the establishment of the Delhi Sultanate (1206-1526) and its expansion into central and south India, the rise of the Mughal Empire, up through religious counter-mobilization by the Hindu Maratha Empire and their eventual defeat at the hands of the British (1674-1818). In total, our dataset identifies over 300 Hindu-Muslim conflict dyads covering the period 1000-1850 A.D. Because all of our historical data come from one source – and a source that was likely biased, as colonial officials wished to highlight India’s violent history and subsequent Pax Britannica – we carefully cross-referenced all 300+ conflicts with a variety of additional secondary source materials, giving us confidence that the information we compiled from the IGI is reliable. Our dataset, coding protocol, cross-referencing protocol, and information on intercoder reliability are all included in the Appendix. 12 Kruijtzer (2009) 4 We examine how the historical incidence of religious conflict in India is related to postcolonial Hindu-Muslim violence, using dependent variables from the colonial era as well as the Varshney-Wilkinson dataset on communal riots from 1950-95. Our main independent variable is Hindu-Muslim conflict “stock,” a measure that captures the cumulative effect of conflict incidents over time. We find that Hindu-Muslim violence in Indian history begins to correlate with contemporary rioting roughly around 1700 A.D. – before the British ruled the subcontinent. Our results are robust to the inclusion of a variety of control variables, different model specifications, and additional checks, including a regression discontinuity design. Our findings therefore cast significant doubt on the widespread view that British colonialism is the original cause of Hindu-Muslim conflict. Notably, however, we also find no evidence that religious violence dates to the earliest waves of Muslim invasions, a rejoinder to Hindu nationalist parties and organizations that have tried to depict Hindus suffering “a thousand years of slavery” under Muslim hegemony, the view of current Indian Prime Minister Narendra Modi.13 In other words, Hindu-Muslim conflict is neither ancient nor a 20th century phenomenon – its roots lie in the often-overlooked precolonial period. We contend that the causal mechanism driving this statistical relationship is ethnic counter-mobilization. Many scholars of Indian politics have pinpointed militant Hindu nationalist groups as a major cause of contemporary riots,14 and we complement this view by showing that the seeds of Hindutva (lit: Hindu-ness) were sowed in the areas where the Mughals and Marathas engaged in extensive conflict – mostly in the modern-day states of Maharashtra and Gujarat. In short, Hindu nationalist groups emerged in 17th century India’s most war-torn regions, became institutionalized in the political system over time, and it is these groups who systematically instigate violence against Muslims today. Our analysis has three broader implications for the study of ethnicity and ethnic violence. We contend that scholars interested in the roots of modern ethnic conflict across the nonwestern world have exaggerated the impact of colonialism. Scholarship that privileges a colonial origins narrative tends to pay inadequate attention to countervailing research on the precolonial period.15 This paper demonstrates the value of a deeper historical approach: in India there is a clear connection between areas that 13 See Shourie (1998) for an example of this kind of Hindu nationalist history. See also Dalrymple (2005) for a broader discussion of the battle over Indian history. 14 Brass (1997; 2003); Varshney (2002); Wilkinson (2004) 15 Lustick (1996); Kreuzer (2010) 5 experienced precolonial Hindu-Muslim conflict and areas that experience contemporary communal riots. We are not suggesting that colonialism was unimportant in the production of modern ethnic violence; rather, our goal is to show that precolonial legacies were often, to paraphrase Marx, the nightmare that weighed on the brains of colonial administrators and the populations they governed. A second implication of our work is the special distinction we draw around the precolonial nature of religious identities and religious conflict. While there are good reasons to suspect that caste or tribal identities in India were deeply structured by colonialism (which we discuss later), neither Hinduism nor Islam are purely western constructs. Religious studies scholarship on Hinduism, for example, routinely describes it as “the world’s oldest religion,” but most social scientists cling to the view that Hinduism became unified only in the early 19th century. A final implication of our study is its focus on ethnic counter-mobilization as a causal mechanism connecting historical and contemporary violence. During the rise of the Marathas, dominant castes used religious identity and symbolism to consolidate the support of the broader Hindu population, leading to severe conflict around Maharashtra and Gujarat – and these areas are the twin strongholds of Hindu nationalism today. In the conclusion, we show how this same mechanism can be applied to other cases. II Colonialism and Ethnic Violence The study of ethnicity and ethnic conflict is tremendously diverse, and therefore it is important to clearly define terms at the outset. Ethnicity is a descent-based social identity. While individuals have multiple identities that they can use to reduce the complexity of the social world, ethnic identities are uniquely powerful because ethnic groups share myths of a common origin, a sense of a common fate, a common culture and symbols, physical similarities, and they face reduced barriers to communication.16 Ethnic violence is violence that occurs largely along ethnic lines, “in which the violence is coded as having been meaningfully oriented in some way to the different ethnicity of the target.” 17 Focusing our discussion on ethnic violence in the nonwestern world, we begin by reviewing scholarship that sees the origins of ethnic groups and conflict in the era of European colonialism, mainly during the 19th and 20th centuries. We then discuss a critique of this view, drawing on the work of 16 17 Brubaker, Loveman, and Stamatov (2004); Hale (2004), (2008). Brubaker and Laitin (1998: 428). 6 scholars who argue that while the colonial period may have reified ethnic groups and entrenched ethnic animosities, both of these developments began to emerge during precolonial times. It is now commonplace for social scientists to point to the impact of colonialism in producing ethnic violence across the globe. As discussed previously, this perspective usually proceeds via three arguments: first, that ethnic boundaries in the precolonial period were fluid. Second, that colonial empires constructed modern ethnic identities through a variety of state-making policies. Third, that colonial administrators instituted self-serving policies of ethnic favoritism and divide-and-rule that gave rise to group conflict. To the extent that scholars of historical ethnic conflict make reference to the precolonial past, it is usually to argue that group identities existed in precolonial societies but were “fluid” or “malleable.” Terence Ranger notes, for example, that pre-colonial African societies “valued custom and continuity but custom was loosely defined and infinitely flexible.” 18 Andreas Wimmer argues more broadly that premodern polities created hierarchical but “genuinely non-ethnic” states.19 As a corollary to this view, scholars also contend that ethnic violence was rare in precolonial times precisely because flexible relations between communities promoted mutual tolerance and harmony.20 Lange has gone so far as to contend that during the precolonial period, ethnic violence itself did not exist.21 Group violence existed, of course, but to call this ethnic violence would be a category error.22 For example, conflict between tribes in Africa could not have been ethnic conflict because these tribes were not exclusivist communities. The advent of colonialism upended this traditional system of ethnic fluidity and cohesion. Much of the classic work on modern ethnic identity formation focuses on European cases and the internal mechanisms of war-making, print capitalism, and rationalization and bureaucratization.23 In the colonial context, however, this process of ethnic identity construction was not indigenous. Rather, it was the colonial state that is taken to have built modern states and constructed modern ethnic groups, largely through 18 Ranger (1983: 247). Notably, Ranger later softened his stance – see Ranger (1993). Wimmer (2002: 2). 20 On Africa, see Reid (2011). A similar view exists on the Indian case regarding HinduMuslim “syncretism”; see Roy (1984), Bayly (1985), Nandy (1988), Thapar (1993) and Hasan and Roy (2005). 21 Lange (2017) 22 Lange (2017: Chapter 2). 23 Anderson (1983); Tilly (1992); Giddens (1991). 19 7 policies of classification and categorization. One of the most commonly hypothesized mechanisms of ethnic identity construction during colonialism is the census. Sudipta Kaviraj, for example, notes a distinction between “fuzzy” communities in precolonial India and “enumerated” communities that came about as a result of colonial data gathering projects.24 Additional studies on Malaysia, the Philippines, and India also point to the power of colonial censuses.25 A passage from Daniel Posner’s work on Africa succinctly showcases the conventional view of colonialism’s impact on ethnic groups: Howard Wolpe shows that the Igbo of Nigeria were a product of colonial boundaries. Terence Ranger shows that the Manyika of Zimbabwe were “created” by missionaries. Crawford Young traces the origins of the Ngala of present-day Congo to Henry Stanley’s misinformed labeling of the people he encountered on his river explorations. Philip Gourevitch shows that the emergence of Hutus and Tutsis as distinct identity groups in Rwanda was a product of Belgian administrative fiat.26 Therefore, it was only once ethnic groups were constructed that ethnic violence came into existence. This violence emerged through multiple pathways. The census, for example, constructed ethnic groups but also engendered ethnic animosities, as rival communities jockeyed for position, often in order to receive patronage from the colonial state.27 Additionally, colonial elites implemented policies of ethnic favoritism and divide-and-rule in order to maintain their power.28 For example, the Belgians in Rwanda favored the minority Tutsi community,29 while the British in Malaya imported Chinese and Indian workers because they believed that Malays were a lazy race.30 These policies led to waves of ethnic conflict in many colonial states in the late 19th and 20th centuries.31 Despite the popularity of the colonial origins literature, other scholars have identified significant flaws with all three of its main arguments. At the 24 Kaviraj (2014). See Charles Hirschman (1987) on Malaysia, Benito Vergara (1995) on the Philippines, and Lieberman and Singh’s (2017) recent study of India. 26 Posner (2003: 127). 27 Lieberman and Singh (2012). 28 Pandey (1990). 29 Hintjens (1999) 30 Shoup (2008: 46) 31 Dwyer and Nettelbeck (2017) 25 8 outset, the assumption that communities from Igbos to Shia’as to Hindus had no clear ideas of group differentiation prior to the 19th century is doubtful, and evidence to the contrary is dismissed as “reverse-projecting” modern identities onto the past.32 But in many cases, ethnic group differentiation clearly preceded colonialism. Richard Reid’s review of how precolonial African history has been marginalized notes that “political scientists and historians alike have been much more excited by what colonialism ‘imagined’ and ‘invented’ than by what already existed and often, indeed, than by the multifaceted question of African agency itself.” Anthony D. Smith has gone the farthest in elaborating this position, arguing that across the history of nonwestern nations, it was often ethnic communities (ethnie) that built states and not the other way around. It should be noted that these scholars rarely contend that ethnic ties are ancient per se, only that they predate the modern colonial era. The colonial origins literature has also been accused of romanticizing group relations prior to the modern period. While some scholars of African politics suggest that tribal conflict only increased with the advent of colonialism, this claim is historically questionable.33 Besley and Reynal-Querol show, in contrast, that conflict in Africa during the period 1400-1700 is correlated with contemporary violence,34 Many scholars of Hindu-Muslim conflict have advanced a similar perspective where India before colonization had a “composite culture” – marked by what Gandhi called sarva dharma sambhava, or equality between religions – in which Hindus and Muslims peacefully coexisted. As we detail in a later section, this view of India’s precolonial golden age is rather idealized.35 The second assumption behind the colonial origins literature is that modern ethnic groups were constructed only in the wake of colonialism. This view essentially contends that Europeans created modernity in the nonwestern world.36 But even a mechanism of identity construction as central as the colonial census is quite problematic. It is rarely acknowledged that censuses were conducted before colonialism,37 and it would be wrong to describe colonial censuses as simply imposed from above – rather, imperial categories 32 Lieberman and Singh (2017: 38). Kasozi, Musisi, and Sejjengo (1994); Reid (2007). 34 Besley and Reynal-Querol (2014). 35 Additionally, this harkening back to the golden age constitutes a political project of India’s postcolonial elites, who tried to downplay historical religious violence in the wake of partition. See Gaborieau (1985) and Bhargava (2010). 36 For example, see Reid (2011) for a discussion of how the precolonial period has been marginalized in African history. See Eaton (2000) for a similar critique of the precolonialcolonial-postcolonial approach to Indian history. 37 See, for example, Peabody’s work (2001) on pre-British censuses in India. 33 9 were formed through a process of engaging with indigenous sources and interacting with local interlocutors. Anastasia Piliavsky’s work, for instance, shows that the supposed colonial-era category of Indian “Criminal Tribes” had a deep-rooted indigenous history.38 The final colonial origins argument places special emphasis on policies of ethnic favoritism and divide-and-rule as the cause of modern ethnic violence. It is well documented that colonial powers implemented certain policies that ostensibly divided ethnic groups, but scholars have called into question whether these groups were already divided prior to colonial rule. If tribal conflict in Africa, for example, was also prevalent in the precolonial period, then subsequent colonial policies may have been intended to reduce rather than increase violence. III The Case of Hindu-Muslim Conflict in India (i) British Rule and Religious Conflict One of the most significant problems in the contemporary Indian Republic is the prevalence of Hindu-Muslim violence, which has, as Steven Wilkinson notes: “...split the Indian state apart once already and has the potential to do so again.” 39 India is a Hindu majority country but is also home to one of the largest and oldest Muslim populations in the world. Since the 1980s, with the rise of the Hindu nationalist Bharatiya Janata Party (BJP), communal riots have increased, leading to the destruction of the Babri Mosque in Ayodhya in 1992 and the 2002 pogroms in Godhra. Since the 2014 election of the BJP, India has been beset by rising sectarian intolerance and anti-Muslim mob lynchings. This conflict between Hindus and Muslims has often been portrayed as a quintessential example of the legacy of colonial divide-and-rule policies. Many scholars have argued that Hindu and Muslim identities were fluid and amorphous in precolonial India. This argument usually begins with basic terminology. The word Hindu is described as originally a geographical term derived from the Sanskrit word sindhu for the Indus river.40 And Hindu is therefore taken to be an inclusive term for those who lived east of the Indus. Similarly, the term Muslim was rarely used in medieval Indian texts; 38 39 40 Piliavsky (2015) Wilkinson (2004: 13). Talbot (1995); Lieberman and Singh (2017). 10 instead, Muslims were called Turushka (Turks), mleccha, or Yavanas, a generic term for outsiders that had previously been used for the Greeks.41 Moreover, Hindu and Muslim communities were hardly monolithic: Hindus were internally divided by caste, and the broad category of Muslim applied to an array of groups, including Turks, Afghans, Arabs, and Persians.42 Due to fluid boundaries between Hindus and Muslims, scholars note that a composite religious culture emerged in India where violence was rare. For example, Truschke shows how Mughal emperors translated Sanskrit texts into Persian, and engaged with Hindu, Buddhist, and Jain philosophers.43 Asim Roy notes the syncretic religious culture that emerged in Bengal.44 The historian Mushirul Hasan summarizes: “The dominant picture of the seventeenth and eighteenth centuries is not of the Hindus and Muslims forming exclusive and antagonistic categories, but of their cooperating in cultural and social affairs.” 45 It is often argued that the advent of British rule turned these fluid, amorphous groups into the modern Hindu and Muslim communities. Paul Brass notes that the colonial state hardened a divide between religious groups.46 Gyanendra Pandey argues that the British tended to view the whole of Indian history through the lens of religious divisions.47 Ayesha Jalal writes that the colonial state viewed the nationalist movement “erroneously...in terms of the great religious divide between Hindus and Muslims.” 48 One of the most-discussed mechanisms of religious identity construction during the British colonial period is the census. The term Hindu, for example, was imposed on a range of diverse groups, flattening out differences between sects like Vaishnavas who worshipped Vishnu and Shaivites who worshipped Shiva. And in the eyes of British administrators, Muslims were considered irreconcilably different from their Hindu counterparts. Once exclusivist Hindu and Muslim identities were constructed, the British also implemented divide-and-rule policies that engendered religious conflict, a view first systematically advanced by Indian nationalist historians.49 Lieberman and Singh’s recent article uses Hindu-Muslim conflict as a case 41 42 43 44 45 46 47 48 49 Laine (2003) Mukhia (1972); Jha (1998); Thapar (1993) Truschke (2016) Roy (1984) Quoted in Ollapally (2008: 26-27) Brass (1974) Pandey (1990) Jalal (1994: 11). Rai (1928); Gopal (1963). 11 study to illustrate how colonial censuses demarcated communities and intensified group conflict.50 Other policies such as the Partition of Bengal in 1905, which split the province along Hindu and Muslim lines, and the Morley-Minto reforms of 1909, which created separate Hindu and Muslim electorates, are often described as driving a wedge between religious communities. (ii) The Precolonial Roots of Hindu-Muslim Conflict In contrast to existing work, we argue that while colonial policies may have hardened ethnic divides, the origins of Hindu-Muslim conflict date to the precolonial period. Our theory of Hindu-Muslim conflict proceeds in two steps. First, we argue – drawing especially on the work of scholars of Indian religions – that Hindu and Muslim identities began to emerge before colonialism. Then, we argue that Hindu-Muslim violence was also prevalent before colonial rule, and that this conflict was vital in constructing religious communities. The roots of modern communal violence in India date specifically to the 17th century conflicts between the Mughal emperor Aurangzeb and the Maratha king Shivaji. When were the Hindu and Muslim communities formed? Returning to terminology is an instructive place to start. While Hindu is routinely described as a purely geographical term, David Lorenzen notes that Muslims born in India (i.e., east of the Indus) were not called and did not call themselves Hindus. Clearly the term referred to more than geography alone.51 Similarly, while Muslims were usually known by a variety of appellations, Harbans Mukhia notes that the often-used term Turk was deployed in medieval literature as “a synonym for Muslim.” 52 The work of the mystic poet Kabir (c. 15th century) details that this term was often used in a clear religious context: The Hindus and the Turks live close together. Each makes fun of the other’s religion (dhamme). One calls the faithful to prayer. The other recites the Vedas. One butchers animals by bleeding, The other cuts (off their heads). Some are called ojhas, others khvajas. Some (read) astrological signs, others fast in Ramadan. Some eat from copper plates, others from pottery. 50 51 52 Lieberman and Singh (2017). Lorenzen (1999: 635). Mukhia (2004: 22). 12 Some practice namaz, others do puja.53 The timeline for the formation of Hindu and Muslim identities is contested, but recent research suggests that it was a gradual process that occurred between the 12th and 16th centuries. Scholars point to the arrival of Islam in north India in the 11th century as, more than anything, helping to form consolidated religious identities.54 While both Hindus and Muslims were heterogeneous groups, their internal differences were dwarfed by cross-ethnic differences. At the most basic level of interaction, there were stark contrasts between Hindus and Muslims that get glossed over by those focusing on the colonial period. Hinduism was a polytheistic and orthoprax religion in which idols were central to ritual. Islam, by contrast, was a monotheistic and orthodox religion that abhorred the practice of idolatry.55 Al-Biruni, a philosopher in the court of Mahmud of Ghazni, famously recognized these differences in the 11th century: They (the Hindus) totally differ from us in religion, as we believe in nothing in which they believe and vice versa...Their fanaticism is directed against those who do not belong to them – against all foreigners. They call them mleccha, i.e. impure, and forbid having any connection with them, be it by marriage or any other kind of relationship, or by sitting, eating, drinking with them, because thereby they think they would be polluted. The Hindus claim to differ from us, and to be something better than we, as we on our side, of course, do vice versa.56 Notably, while Al-Biruni highlighted stark differences over religion, his depiction of the caste system emphasized its fluidity, at least outside the elites: “Between the latter two classes [castes] there is no very great distance. Much, however, as these classes differ from each other, they live together in the same towns and villages, mixed together in the same houses and lodgings.” The later work of vernacular poets like Eknath, Kabir, Tukaram, and Ramdas also contains ample references to religion.57 These authors are especially important because they came from non-Brahmin castes, signaling 53 Lorenzen (1999: 651). Talbot (1995); Lorenzen (1999). 55 This profound difference between monotheistic and polytheistic faiths has been noted as being at the center of historical conflict in other parts of the world. See Schwartz (1998); Assmann (2009). 56 Gaborieau (1985: 7). 57 Michaels (2004). 54 13 that Hindu and Muslim communities before the British were not simply the impressions of elites.58 Kabir, for example, sarcastically assails both Hindu and Muslim beliefs, indicating the salience of two religious groups.59 Andrew Nicholson’s work – which focuses on the philosopher Vijnanabhiksu (c. 16th century), who attempted to unify various diverse schools of Hindu philosophy – notes that by the late medieval period, it “...became almost universally accepted that there was a fixed group of Indian philosophies in basic agreement with one another and standing together against Buddhism and Jainism.” 60 There is also significant evidence that Hindu-Muslim conflict predates colonial rule. We would go further and note that it was religious rivalry and violence that helped demarcate Hindu and Muslim communities.61 John F. Richards finds that the earliest Islamic forces in India “appealed regularly to Muslim militancy in the jihad or holy war against the idolatrous Hindus of the subcontinent.” 62 Al-Biruni described the conquests of Mahmud of Ghazni as “wonderful exploits, by which the Hindus became like atoms of dust scattered in all directions.” 63 On the other side, Mukhia notes that during the time of the Mughal emperor Akbar (r. 1556-1605), “...the eminent Islamic theologian Shaikh Ahmad Sirhindi complained that ‘The infidels are demolishing mosques and are building their own places of worship in their stead. In Thanesar in the Kurukhet [Kurukshetra] tank there was a mosque and the shrine of a saint. Both these have been destroyed by the infidels and in their place they now have a big temple.” ’64 While these early clashes are important, we would date the roots of HinduMuslim violence to the Mughal-Maratha conflicts of the 17th century. For most of the Mughal period, emperors embraced ecumenical policies in order to effectively govern their large and diverse realm. For example, Akbar famously rescinded the jizyah, and Thomas Roe, the British ambassador to the Mughal Court, noted that during the reign of Jahangir (r. 1605-1627): “...all sorts of religions are wellcome and free...” 65 All of this changed during the reign of the sixth and last great Mughal emperor. Aurangzeb is a controversial figure, and although he was hardly the Islamist tyrant 58 Lorenzen (1999: 648). Gottschalk (2000: 20). 60 Nicholson (2010: 2-3). 61 Barth (1969). 62 Richards (1993: 2). 63 Pollock (1993: 284). Al-Biruni may have been referring specifically to a Brahmin exodus from the Punjab – see Wink (1997: 123). 64 Mukhia (2004: 26-27). 65 Mukhia (2004: 20). 59 14 portrayed with fury by Hindu nationalists,66 it is undeniable that several new policies instituted during his reign served to embitter relations between Hindus and Muslims. Notably, Aurangzeb broke from the ecumenical policies of his forebears by increasing Hindu customs duties; shutting down Hindu fairs; restricting practices associated with celebrating the festivals of Diwali and Holi; and, most important, reimposing the jizyah.67 These policies, instituted in the late 17th century, fueled the fires of Hindu-Muslim discord at a time when the British were a minor power trying to consolidate small trading settlements in far-flung corners of the Mughal Empire. For example, Niccolo Manucci, an Italian writer working in the Mughal court, noted in the 17th century: In this realm of India, although King Aurangzeb destroyed numerous temples, there does not thereby fail to be many left at different places, both in his empire and in the territories subject to the tributary Princes. All of them are thronged with worshippers; even those that are destroyed are still venerated by the Hindus and visited for the offering of alms.68 Aurangzeb greatly expanded his empire through relentless military conquest, especially in the Deccan region of central and south India. Shivaji was a king of the local Maratha warrior clan who had at times been an ally of the Mughals, but by the late 17th century had turned against them and succeeded in carving out an independent kingdom. Shivaji’s nascent Maratha Empire, in contrast to earlier ostensibly Hindu polities, was more self-conscious in its use of religion as a form of legitimation, as well as a form of resistance to Islamic rule. For example, Shivaji’s letters to Aurangzeb signaled his disapproval of the emperor’s bigoted policies against Hindus. In one, he wrote about his opposition to the jizyah: The fire of rebellion, born of the torture of innocents, can burn the whole kingdom faster than any fire. The Emperor therefore should not discriminate against any religious creed and oppress them... However, if Hindu people are subjected to misery, your empire will be reduced to ashes in the fire of their anger.69 A common refrain is to downplay this violence as about political and not religious goals, but this ignores the fact that religion and politics are inherently fused in both Hinduism and Islam. Kingship is a central concept 66 67 68 69 Brown (2007). On Aurangzeb, see Sarkar (1912); Chandra (1969); Mukhia (2004). Mukhia (2004: 25). Pansare (2015: 1973). 15 in Hinduism,70 and in Muslim kingdoms the ulama (Islamic scholars) was central to governance.71 Hindu and Muslim polities in medieval India were therefore sectarian by nature. Marc Gaborieau writes that medieval states “...were as a rule religious states. The religious community of the ruling oligarchy (even if it was in a minority) had a privileged status; Muslims were privileged in Muslim states and Hindus in Hindu states.” 72 Furthermore, we can call the Mughal and Maratha empires Muslim and Hindu empires because they self-consciously used the language of religion to justify and legitimize their policies. Saqi Mustad Khan, employed in Aurangzeb’s court and author of an authoritative account of the emperor’s life, Maasir-i-Alamgiri, described his infamous decision to reimpose the jizyah as follows: As all the aims of the religious Emperor were directed to the spreading of the law of Islam and the overthrow of the practice of the infidels, he issued orders to the...officers that from Wednesday, the 2nd April, 1679... in obedience to the Qur’anic injunction “till they pay commutation money (jizyah) with the hand in humility” and in agreement with the canonical traditions, jizyah should be collected from the infidels...of the capital and the provinces.73 On the other side, Shivaji drew explicitly on ancient notions of Hindu kingship and was coronated in a ceremony in which thousands of Brahmin priests recited Vedic texts.74 It is important to take these references to religion seriously; as Anna Grzymala-Busse argues: “Such statements may be instrumental, rather than sincere – but they are not said without a reason. The instrumental use of religious rhetoric nonetheless presupposes a willing audience, and an expectation that such claims will resonate.” 75 All of this evidence indicates that Hindu and Muslim identities began to emerge before the colonial age, as did conflict between these groups. This conflict was also essential to demarcating religious boundaries. This is not to say that British rule had no effect on further hardening communal identities in the 19th and 20th centuries. However, our narrative in this article questions whether the British actually intended to divide Hindus and Muslims or whether their real goal was to separate them and minimize 70 71 72 73 74 75 Fuller (2004: Chapter 5). Mukhia (2004). Gaborieau (1985: 9). Chandra (1969: 323). Wolpert (2009: 169); Kruijtzer (2009). Grzymala-Busse (2016: 335). 16 violence. This alternative view is supported by Francis Robinson, who writes that the British government made “no deliberate attempt to foster communal hostility; indeed the aim was to avoid it.” 76 Peter Hardy further notes that “To balance and rule, not divide and rule, was the instinctive British approach to politics in India.” 77 IV Empirical Analysis (i) Data on Historical Hindu-Muslim Conflict We test our theory of the precolonial roots of contemporary Hindu-Muslim violence using a new dataset of historical religious conflict i n I ndia. The underlying source for our analysis is the Imperial Gazetteer of India, a 26-volume British history of the subcontinent.78 This encyclopedia contains detailed geographic, demographic, economic, and historical sections for every district in India, and many district entries contain information about instances of religious conflict. T o b e c lear, t he I GI i s n ot a primary source; rather, it is a compilation of primary sources, secondary sources, oral histories, and other historical material. Additionally, the IGI mainly codes elite-level politics. But the IGI is a serious work of history, and it has been used extensively by historians and social scientists of India.79 The IGI is comparable to the Gazetteer of the Persian Gulf, Oman and Central Arabia, written by John Gordon Lorimer,80 which is often used in social science today among those who study the Middle East. We employed two research assistants to code incidents of violence mentioned in the IGI. The first r esearch a ssistant c oded a s ubset o f t he I GI without any guidelines. On the basis of this work, we developed a detailed coding protocol, which was then used by a second research assistant to code all 26 volumes. We used multiple coders to ensure intercoder reliability.81 Our research assistants entered the following information for each instance of conflict: location, an event category (“war,” “battle,” “raid,” etc.), 76 Robinson (1974: 348-49) Hardy (1972: 89) 78 Meyer et al. (1909). 79 The IGI has been referenced in the work of historians like Subrahmanyam (2002), as well as development economists like Iyer (2010). 80 Lorimer (1908). 81 Krippendorff (1980). 77 17 year, attacker polity, defender polity, attacker religion, and defender religion. To illustrate how the coding process worked, we provide a basic example from the IGI for Anantapur District in the Madras Presidency of South India. The IGI “History” section for Anantapur is shown below:82 Figure 1: Excerpt from the Imperial Gazetteer Our dataset then recorded the following incidents of violence and corresponding information: 82 Meyer et al. (1909 Vol. 5: 339). 18