Archival ethnography and ethnography of archiving: Towards an anthropology of riot inquiry commission reports in postcolonial India

ABSTRACT This paper examines the challenges and possibilities of combining archival and ethnographic methods in the field of ‘communal’ violence studies in India. Drawing insights from debates among historians and anthropologists on the multifarious interactions between archives and ethnography and reflecting on the empirical case of persistent violence between Muslims and Christians in southern India, it argues for a creative synthesis of these two modes of inquiry for an adequate understanding of ‘communal’ violence and riot inquiry commissions in India. First, the paper critiques how colonial and postcolonial Indian archival reports problematically inscribe violence between any religious communities (such as Muslims and Christians) in the same narrative as the predominant case of Hindu-Muslim conflict. Second, it illuminates how archival ethnography can be an effective way of studying violence between religious communities and thus transcend conventional disciplinary boundaries. Finally, the paper introduces a nuanced approach, called ‘ethnography of archiving’, to detail the judicial and nonjudicial discourses and bureaucratic manoeuvring involved in the creation of an archival report, thereby unravelling the power relations, mediating processes, manipulations and bureaucratic performances that make commission reports problematic even today.


Introduction
The complex relationship between archives and ethnography has been a central concern of social science for the past few decades, and there is now an exciting body of literature, especially from historians and anthropologists, that addresses the limits, possibilities and necessities of using them together to study a range of issues across times and locations (e.g. Axel 2002;Bennett 2014;Camaroff and Camaroff 1992;Carminati 2019;Cohn 1987;Cunha 2006;van der Veer 2001;Zeitlyn 2012). This paper examines the possibilities and challenges of such a creative synthesis in the field of what is often termed 'communal violence', where it has been little used so far. 1 Narrowly confined to the respective disciplinary poles of history and anthropology, archival research and ethnographic fieldwork are invariably postulated as opposing epistemological and empirical modes of inquiring into 'communal' violence in South Asia. 2 Although some works have brought both of these together to study violence, a conscious effort to elaborate the theoretical and methodological possibilities and challenges of their innovative fusing has not yet been attempted. 3 Archives have been almost exclusively used by historians to construct the past of 'communal' riots in India (e.g. Chandra 1984;Pandey 1991;Seal 1973). They have also been critiqued by scholars, especially how 'communal' violence, including its participants, is inscribed in problematic ways in documents created since the colonial era (Baxi 2007;Das et al. 2000Das et al. , 2001Hansen 2001;Mehta 2002Mehta , 2007Pandey 1991). Moreover, ethnographic works which focus on the everyday and agential dimensions of 'communal' violence (e.g. Arif 2016; Chatterji and Mehta 2007;Das et al. 2000Das et al. , 2001Mehta 2002Mehta , 2007 have emerged to challenge the macro, causal and objective modes of analysis propounded primarily by historians. Central to the difference between these two major strands of studying riots is a series of conventional binaries in social science: interpretation versus explanation, objectivity versus subjectivity, macro versus micro. Going beyond these dualisms, this paper brings forth the intricacies involved in a creative combination of archives and ethnography for a better anthropological understanding of riots and their inscription by the inquiry commissions set up by the government after riots in postcolonial India. 4 How does the archive's function as a source of governmental power through classification and surveillance inform archival reports on violence between religious communities in India? Can we use riot archives as ethnographic sites, even though they are problematic in many senses? In other words, how can we understand the 'experience' of the event, 'be there' ethnographically and listen to the 'voice' of the actors involved in riots from archival reports, surpassing their problematic assumptions? Finally, what lessons can we learn if we undertake the 'ethnography of archiving', tracking the very process of creating reports of violence? To put it differently, what insights can anthropologists glean from following the courtroom procedures, judicial discourse and everyday proceedings of a commission of inquiry in a given context of violence, which later end up in a report to be kept in an archive? By investigating the case of Muslim-Christian violence in Kerala, this paper throws new methodological and theoretical light on the potential combination of ethnography and archives in studying riots in India.

A case of noncommunalized violence
The empirical context of this paper is the PhD research I carried out on the long history of violence between Muslims and Christians in the coastal villages of Thiruvananthapuram district (also still called Trivandrum, its colonial name) in Kerala, the state at India's southwestern edge. The trigger for the investigation was the basic assumption of all 'communal' violence studies and archival reports in India that 'mobilization of religious identity' is a precondition for violence between members of different religious communities in India (e.g. Bayly 1985;Brass 1996;Chandra 1984;Das 1990;Dumont 1970;Kakar 1996;Nandy 1999;van der Veer 1994;Varshney 2002). 5 This assumption, rooted in orientalist and colonialist notions about South Asian societies, has its origin in administrative, discursive, institutional and academic practices that began in colonial times (Pandey 1991;Punathil 2019;Sarkar 1999). Riot reports and social science literature in postcolonial India continue to reinforce the idea, directly or indirectly. But as Punathil has shown (2019), a historical anthropological analysis of sociospatial dynamics, it is the reification of community identities, rather than communalization, that explains the occurrence of violence between Muslims and Christians over the course of a century in this contested area. As Brubaker (2002) describes, reification of group identity is a process whereby that identity becomes substantial and fixed; it is a process of 'thingification'. Violence may happen between individuals from two communities when situations, events or activities reify their groupness. Reification can take place without a group's being communalized or religious identity mobilized; mobilization is only one way of reifying identity in escalating attacks between groups.
Thiruvananthapuram saw three phases of violence. The first, formative stage occurred between 1900 and 1950 as Mukkuvar Christians seeking better livelihoods settled in the coastal villages and came into conflict with the numerous and economically powerful local Muslims, called Marakkayars, over land, fish and other resources. The interplay of identity consolidation, local hierarchies and fights over space led to the regular eruption of small but violent encounters throughout the region. The second phase played out against the background of India's enormous post-Independence structural transformations: between 1950 and 1980, Mukkuvars, aided by the church, adopted mechanized fishing practices and quickly eclipsed the Marakkayars in economic power and status, resulting in escalating, lethal violence as their social transformations were deeply implicated in the contested space. It was not religion but the everyday dynamics of community life that reified identities here. A major shift in the nature of violence characterized the third phase, after the 1980s. As Muslims withdrew from fishing, conflict between these two communities declined. However, it has not disappeared, as violence continues to erupt, including two major riots in 1995 and an infamous incident of police killing Muslims in 2009. The state was the leading player in this phase, intruding into this contested space by reifying the identities of Muslims and Christians. Because the violence I investigate has a long history, it raises the methodological challenge of connecting the past and the present and thus combining the practices of history and anthropology in a way that does not fall prey to prevailing binaries. In this paper, I will engage with three broad methodological issues that emerged from my work in this context, broadly classified as critique of archives, archival ethnography and ethnography of archiving.

The critique of archival riot reports
There are several government reports on violence between Muslims and Christians in Thiruvananthapuram coastal villages. While a few (including letters) are from colonial times, most concern postcolonial incidents, the latest occurring in 1995. Entering the archival world of riots confronts one with several challenges. A few scholars have already criticized Indian riot reports as heavily influence by colonialist sources (Hansen 2001;Mehta 2002;Pandey 1991), generally based on the criticism of the archive as an instrument of those who govern (Trundle and Kaplonski 2011). The most powerful critique comes from those who advocate Derrida's (1995) idea that archives represent the modern powers of classification, representation and surveillance of subjects. Foucault (1977) also sees archives as hegemonic ways of thought, modes of colonization and control of citizens: expressions of government's control of its subjects. In this view, archives are sites where authority or social order is implemented (Cunha 2006;Zeitlyn 2012). Indeed, the material they conserve can be closely linked to the construction of colonial knowledge and the practices of colonial agents (Cunha 2006). However, the archive is not merely a store of governing data, with its statistics, enumeration and classifications, but also a corpus of power and authority, with orientalist notions influencing colonial ethnographic practices (Appadurai 1981;Asad 2002;Bennett 2014;van der Veer 2001). Therefore, both ethnography and statistics were political disciplines in colonial administration (Axel 2002). In a nutshell, the problem with archives is that they are produced through interactions among diverse mediators of colonialism and colonized subjects. More important, scholars such as Brian Keith Axel (2002) have argued that even archival documents relating to precolonial history have never been free from colonialism, because archival knowledge must be seen as something that emerged from colonial processes and the conditions that allowed for the production of historical knowledge. In the same vein, colonialism and its production and dissemination of knowledge through archives continue to be very influential in the knowledge production of postcolonial states. By analysing group violence and its representation in postcolonial India, Arjun Appadurai (1993) has shown the precise and distinctive link between colonial practices of enumeration and classification and its present-day manifestations. Thus, archives as sources for researching 'communal' violence in India are charged on several counts (Mehta 2002;Pandey 1991). The communalism discourse in South Asia goes back to the British administrative, institutional and academic discourse which essentializes Hindus and Muslims as inherently antagonistic and views violence between them as a fundamental feature of Indian society (Amin 2002;Ansari 2005;Mehta 2002;Pandey 1991;Punathil 2019). During the colonial period, any violence between the members of these communities was viewed through the reductive lens of the 'communal'. Pandey (1991) has argued that this perception itself contributed to the rise of violence between Muslims and Hindus. And not just colonialists, but even nationalists in India have been made to believe that communalism is an essential feature of their society and that religion is at the heart of violence in their country. Pandey (1991) quotes numerous examples to illustrate how colonial discourse reduced any violent incident to 'communal riots' by characterizing participants as 'passionate' and the riot as an unavoidable result of the age-old enmity between Hindus and Muslims (see also Baxi 2007). Basing research on such archives therefore risks a reductive reading of violence in India.
Interestingly, colonial-era conflicts between Muslims and Christians were described with the same kind of languageas in, for example, the official letter including the report titled Riots between Muhammadans and Christians of Kollencode, 1939: The [Kollencode] Muhammadans appear to be the aggressors and they seem to have been set up by Poovar 6 Muslims who want to establish a custom to stop music before mosque. I am asking the District Superintendent of the Police, Trivandrum to watch the situation carefully and see that such outbursts of fanaticism are not allowed to take place. 7 Colonial usages appear throughout this report, such as 'Muhammadans' for Muslimswhom, moreover, it characterizes as 'aggressors' given to 'outbursts of fanaticism'. Viswanathan's (1998) study on colonial power and religious conversion in India eloquently demonstrates how such usages have played an important role in constructing distinctive religious identities which became stereotypes that persist to the present day, such as those of Muslims. Such categorizations were uncritically accepted and reiterated by native rulers, administrators, Christian missionaries and even nationalist leaders. A particularly virulent view of Islam appears in a report on the census of Travancore, the princely state whose capital was Thiruvananthapuram: Muhammedanism is called the 'great antagonistic creed', antagonistic I think to Christianity and to every other religion. Intolerance is its chief characteristic, and as a necessary consequence we find that the Muhammadan races all over the world are fanatical and fierce. The bigotry of the creed breeds 'Faqirism' and stifles philosophical thought and enquiry. A religious schism among them is sufficient to estrange the feelings of their kith and kin from schismatics. To the schismatics the term 'wanders from the truth' is applied, the faithful adherents alone being entitled to the name of Muslims (Moslem) or the Believers. The outside worlds are called infidels or kafirs, broad term covering within it Christians, Jews, Hindus, Buddhists and all other religions. Against the infidels or the kafirs the free use of the sword is sanctioned by the Koran … . Ignorance is their great curse and to this cause may be attributed the immense influence exercised to this day by their itinerant 'Tangals' or priests. The Muhammadan classes furnish no converts to Christianity. (Nagam Aiya 1884, 138) This representation of Muslims as dangerous 'others' is both a religious and a racial classification (Kapila 2007;Viswanathan 1998) and reinforces orientalist notions about Muslims in India.
Are such problematic representations limited to governmental and administrative practices? Unfortunately, institutional separation of archives and ethnography is difficult historically, since much of the archives produced in the colonial period consists of ethnographic descriptions by colonial anthropologists, revealing the politics of colonial power and knowledge production (Bennett 2014;Dirks 1987). Samuel Mateer, a colonial missionary, described the character of the Trivandrum 'Muhammedans' at length (interestingly, he also refers to Vizhinjam, discussed below): So ignorant and heathenish are the Muhammadans of Travancore, that they are pronounced by those of East Coast 'worthless'. They do not allow people of other castes or religions to eat with them. Few go to Mecca on pilgrimage. Sickness is attributed to the agency of demons, wherefore some secretly send gifts to the devil temples; many attend the idol procession at [the] Arattu. 8 Many give their children in infancy. Though forbidden to sell arrack or opium, some do this secretly. As education is low, so crime, it is admitted, is excessive amongst them. Little moral discipline is exercised, though they may be excluded from the mosque, and others may refuse to speak or hold intercourse with them or give them fire, for adultery with heathen women, for disobedience, or drunkenness. (Mateer 1883, 74) Viswanathan rightly notes that the 'Muslim' in British census reports is characterized by an ambivalent identity that fits the colonial project. Such classifications 'create an overarching narrative plot in which race, caste, and religion override the self-definitions of the groups being enumerated' (Viswanathan 1998, 157). The above description, with its references to 'gifts to the devil temples' and 'idol procession', illustrates Viswanathan's argument that 'British ethnographic data, census reports, and commissioned surveys … placed the [Indian] Muslims closer in racial features, behavior, habits, and customs to other native inhabitants of India, including Hindus', than to Muslims elsewhere (1998,157). But this fluid, mercurial status meant that Muslims in India were seen as neither foreigners nor fully Hinduized, thus denying them any unity of religious belief or identity with Muslims elsewhere.
Colonial sources continue to influence postcolonial studies of riots too, as social scientists uncritically accept their constructions of communalism and other generalizations about Hindu-Muslim relationships in India (Amin 2002;Baxi 2007;Kakar 1996;Mayaram 1997;Mehta 2002Mehta , 2007Varshney 2002). Anthropologists have particularly emphasized the trap of objectivism and causality inherent in this approach, which ignores everyday micro realities and the voices of the subjects of violence (Baxi 2007;Chatterji and Mehta 2007). Scholars have also called attention to how Hindu-Muslim violence is still being inscribed in the same way that colonizers wrote about it (Baxi 2007;Das et al. 2000;Hansen 2001;Mehta 2007;Pandey 1991). 9 These studies are largely based on reports on Hindu-Muslim violence, especially in north India, but, as this paper shows, violence between local groups of Muslims and Christians is also described in ways typical of Hindu-Muslim 'communal' violence.
The preliminary proclamations of the reports on Thiruvananthapuram riots have a common outline of their responsibilities: each report is a reaction to the petitions of the victims and to the demands for justice of a larger public, including civil right activists, as well as a promise of neutral and rational investigation carried out by a state-appointed commission. However, these introductory remarks already characterize the violence as essentially 'communal', precisely because it is between communities with strong religious identities, albeit Muslims and Christians instead of the expected Muslims and Hindus. The Gopalakrishna Pillai Commission report states that the government of Kerala appointed the commission to enquire into the circumstances and allied matters [that] led to [the] series of communal clashes [which] occurred at Vizhinjam near Thiruvananthapuram on 14-05-1995 and 10-07-1995, which is a matter of public importance. (Government of Kerala 1998, title page) It was specifically to determine (1) The facts about and the circumstances which led to the communal clashes … ; (2) Whether there was any lapse on the part of police in taking timely action and in dealing with the situation; (3) Such other matters as are incidental to and arising out of the above; and (4) To suggest suitable measures to avoid such unfortunate incidents in future (Government of Kerala 1998, 1).
Like some others, the title of this report includes a predetermined cause: Report of the Commission of Inquiry Regarding Communal Clashes at Vizhinjam, Thiruvananthapuram.
The report repeatedly makes clear its assumption that religious communities are inherently violent, stating not only that 'Vizhinjam is a highly tense area where riots are likely to spread all of a sudden. All authorities concerned know about this fact' but also that 'Vizhinjam is considered to be a communally volatile area by the authorities concerned' (Government of Kerala 1998, 4). Even fatal shootings by police are justified by pointing to the 'violent nature' of the victims and the danger they would have represented had the police not acted. The report on the 1963 incident at Panathura thus absolves them: In spite of the best efforts, it was not possible to disperse aggressive Christian fishermen; the warnings of policemen were ignored by the infuriated mob and finally police Superintendent had to order fire. Each round of shots was fired only when they found that the previous shot had no effect on the mob and it had not dispersed the crowd. (Government of Kerala 1963, 49) The report on the police shooting in Valiathura in 1982, which resulted in one death and many injuries, justified the firing by describing the conditions that led to the violence in terms typical of many other reports: The small of group of people who started the demonstration in a peaceful manner swelled into a big crowd and gradually grew violent and started attacking the police by stone throw, [and] the police had to resort to lathicharge first when many of the police men got injured by stone throw and finally had to resort to firing when they were cornered by more severe and intense attack. (Government of Kerala 1984, 92-93) Most of the riot reports end with the statement 'Therefore the police firing was absolutely necessary'.

Archival ethnography in riot reports
The critique of riot reports does not negate the potential for 'archival ethnography'. This idea gains its strength from the relationship between history and anthropology, whose conventional separation, now being challenged, was a consequence of 'positive empiricism' in Western academia (Asad 2002;Axel 2002;Carminati 2019;Cohn 1987;Dirks 1987). Many decades ago, Bernard Cohn (1987, 2) called for researchers to 'treat the materials of history the way an anthropologist treats his field notes' and asserted that it is not disciplinary differences but theoretical considerations that should determine the nature of social science inquiry. Archives thus become potential sites of ethnography when we attend not merely to their content but also to their form and its relationship to that content (Chakrabarty 2004). Talal Asad (2002) argues that ethnography alone may not accurately or adequately represent an experience, so a consideration of the historical context and structural background must always accompany it. As he notes, the ethnographer's focus on fieldwork means that 'many spatio-temporal complexities and variations [are] excluded from the object of study because they [are] not directly observable in the field' (68). Archival ethnography can fill in some of these gaps, offering rich, historical anthropology. Thus, researchers must not view archives merely as the material of another discipline, and must overcome the tendency to see archival labour as a mere extractive enterprise (Carminati 2019;Marcus 1984). In other words, the use of archives demands a conscious effort and creative dialogue with their data to facilitate accurate anthropological readings. Evans-Pritchard (1962, 50) rightly mentioned that 'anthropologists have tended to be uncritical in their use of documentary sources'. A more sophisticated approach would avoid the problem of invoking fragmentary archival sources to confirm colonial notions.
I offer an example of archival ethnography with reference to the spatiality of violence to show how records can enable an ethnographic reading. Space has been an important site of contestation between Muslims and Christians in Thiruvananthapuram, figuring as a key issue in all three phases of violence there. An archival source, the official report on a 1939 incident sent by the first class magistrate of Padmanabhapuram to the Trivandrum district magistrate, offers an important clue about this history in the first half of the twentieth century: There is a Christian church and a Mohammedan mosque at the place, both situated near the sea-side. The mosque is situated nearly a furlong to the east of the church. There is a footpath going eastwards from the church and to the northern side of the mosque and the mosque compound. The cemetery belonging to the church is situated about seven furlongs east from the church and the way to the cemetery is through the northern precincts of the mosque. The body of the diseased was brought to the church after conducting the usual prayers. The corpse was taken along the above said sandy footpath lying to the northern side of the mosque for burial in the Christian cemetery. It is stated that ordinarily corpses are taken for burial from the church through the beach on the southern side of the mosque. Recently for some days, there was a large catch of fish from the sea and the fish were exposed for drying on the beach near the church all along the coast, but for the past 3-4 days there has been no proper sunshine, leaving the fish putrid, and there was a bad stink all along the coast. It is stated that the body in question was therefore taken along the said footpath near the mosque lest it should become desecrated by walking along the sea shore covered with putrid fish. 10 The letter goes on to describe the riot: At about 2.30 pm on 22.03-39 when the burial party numbering about 100 reached the North-Western limit of the mosque compound, some Mohammedans of the locality, numbering about fifty armed with sticks, came and intercepted the burial party and prevented them from proceeding further. It is stated that sticks were freely used by the assaulters and three of the coffin bearers sustained injuries. The funeral party along with the priest returned to the church leaving the coffin at the very place they were intercepted by the assailants. 11 A letter from the district magistrate to the chief secretary states: It is not for me to say anything at this juncture regarding the right of the Christians to take their dead bodies through the path way near the Mohammedans' mosque. The Mohammedans of the locality told me that the footpath in question is part of the registered holding belonging to the mosque. 12 This colonial case shows how a specific spatial and social boundary has historically been a bone of contention in this coastal belt and how localities in the area have generally become 'contested spaces' between Muslims and Christians. This story, narrated in documents about the distant past, offers a crucial link to the insights I gained through my fieldwork, on contestations over space, community boundaries and everyday life in this region. 13 The importance of spatial dynamics in escalating enmity and violence is something I first learned from doing ethnography: the meaning of space in the everyday lives of these communities and their peculiar relations to social boundaries and conflict are observable at the field site. However, archival information on spatial dynamics is another important thread in weaving the historical story of this violence.
Archival ethnography emphasizes that a new interpretation of the archive is possible if the particular circumstances of the construction of its 'voices' are taken as an object of analysis (Camaroff and Camaroff 1992 ). Camaroff and Camaroff (1992) say that if we 'listen' to such voices 'speaking', we can engage in 'dialogue' with the documents, analysing archives as sources that are assembled and maintained by people, social groups and institutions. The knowledge contained in archives must be understood as systems of personal statements, partial truths and historically and culturally formed interpretations subject to rereading and new explanations (Bennett 2014;Camaroff and Camaroff 1992;Cunha 2006). This conscious exercise will transform an archive, seen as an entity invested with power, into a site of ethnography where the researcher can enter into a mindful relationship with the ideas within (Carminati 2019;Osterweil 2013). In this sense, ethnography is continually interpretative, irrespective of the field it informs, interrogating and disrupting 'established' truths in archival materials (Cunha 2006;Korom 2001).
Although archives of riot reports are problematic in many ways, they are not devoid of the voices of subjects. All commission reports are filled with the narratives of victims and witnesses, alongside those of state officials and members of various political organizations. Indeed, a government document on violence cannot be prepared without the testimony of the groups involved in the violence or before the judicial process and court trials have been completed. In fact, commission reports offer 'thick description', with multilayered stories and contestations (Urla and Helepololei 2014;Viswanath 2015). However, a careful analysis of commission reports on riots, especially postcolonial ones, shows that they are discursively constituted through various classificatory practices. All the reports on violence between Muslims and Christians in the context of my investigation, for example, share a common feature, which I call 'party formation': assigning the groups and individual actors involved to parties, such as perpetrators, victims, state officials. The inquiry that precedes the riot report begins by classifying the parties and then has a 'hearing' for each, with a limited number, usually a few people, who speak for the whole group in describing what happened. Likewise, all reports begin with the details of the parties involved (labelled A, B, C, etc.).
To take one example, let us return to the government report on the 1995 Vizhinjam riots. The violence began on May 14, when Christians killed two Muslims. The way that Muslims and Christians narrated the incident offers interesting clues about the relationship between subject formation within archive and its connection with the world, where two communities are in an antagonistic relationship. The report states that there were nine parties involved, further sorted into three major groups: police officials, Muslims and Christians. The police party was represented by the district superintendent and the district collector, while other parties belonged to the Muslim Juma-ath in Thekkumbhagam and the parish council of St Mary's Church. The narratives of the various parties, especially the community members involved in the violence, constitute the voices of the subjects in this document, and their claims and counterclaims are its ethnographic site.
'Noman's land' repeatedly appears in the narratives as a site of contestation: a small piece of land lying between the territories of the Muslims (to the West) and the Christians (to the East) in Vizhinjam that was established as a neutral space by state officials after a riot in 1972. It acts a boundary between these two spatially segregated communities, and neither group has any right over this land. As the report states: From the evidence in this Inquiry it can be seen that the underlying factor which led to the clashes on 10-7-1995 was the dispute in respect of the right regarding the user of the Noman's area … . So Noman's area is to be considered as the most sensitive area in Vizhinjam, because it is around this area that the members of both the communities pick up quarrels. (Government of Kerala 1998, 37) Now let us see how the agency of both communities is asserted in the document. According to party H, a Christian representative named Johnson: two persons belonging to the Christian community came from Pulluvila in a canoe [and entered the 'Noman's land'] and there was an altercation with another man belonging to [the] Muslim community and because of that Muslims become infuriated and took quarrels, used abusive language and attacked the Christians. Subsequently, they threw bombs collectively and fishing equipments were burnt by pouring kerosene. (Government of Kerala 1998, 13) Party F, Isahak, a representative of the Muslims involved, stated that he saw about ten thousand Christians reaching to the Noman's land with lethal weapons like wooden paddles, sticks etc. in their hands. Some of them were uttering "[chonone ellam kollleda, avante okke ellam kathikkeda]" [kill all Chonones (i.e. Muslims) and set fire to all their belongings]. He was terribly frightened and the Christians crossed the Noman's land and proceeded further towards the west … . Without any provocation the police opened fire towards the west [the Muslim side] as per the directions of the Christians. (Government of Kerala 1998, 11) While these narratives are offered to justify the actions of each speaker's reified group, they also give interesting anthropological insights into the volatile sense of space among both communities and its role in leading to violent conflicts. 'Noman's land', a territorial marker established by the state, is a contested terrain for peoples living in an enclosed space, an intricate boundary that must be continuously negotiated in the everyday lives of these communities.
My anthropological reconstruction of the violence here does not merely rely on ethnographic narratives from the field but is also enabled by the discovery of the sense of everyday life, of space and identity, that actively plays out in the voices of the subjects in the riot reports. There are numerous such sites in these documents, even though the articulations are often fashioned in response to the circumstances after the violence. The contesting claims in riot archives are precisely what make them interesting ethnographic sites and fertile sources for synthesis with conventional anthropological fieldwork.

Ethnography of archiving
As is clear from the evidence presented so far, a neat separation of archive and ethnography is counterproductive, if not futile, in contemporary research. I have specifically shown how such a segregation would obstruct an adequate analysis of riots between members of different religious communities in postcolonial societies like India and how archival ethnography can help us transcend such conventional boundaries. But what can we learn from an ethnography of archiving itself? How can it help us understand the making of an archive for an event such as a riot if we track the very process of archiving? In other words, what insights can an anthropologist glean by following the courtroom procedures, judicial discourse and everyday proceedings of a commission of inquiry in a given context of violence, which later form part of the official report on the incident, to be kept in an archive? To suggest some answers to these questions, this section analyses the judicial inquiry into an instance of Muslim-Christian violence and the subsequent lethal police shooting of six Muslims in 2009 in Beemapalli, another Thiruvananthapuram coastal village.
On the afternoon of May 16, some inhabitants of the neighbouring Muslim Beemapalli and Christian Cheriyathura got into a fight over a monetary loan from a Marakkayar to a Mukkuvar. The tension rose so high that both the district collector and a sub-collector were sent to the area and instructed the police to maintain law and order. The next afternoon, however, both communities mobilized in large numbers, and a large police force was sent to the boundary between Beemapalli and Cheriyathura. The official report states that a fight broke out and the police, trying to maintain control, shot at the Marakkayars, killing six and injuring several others. I attended the resulting court and bureaucratic proceedings, ethnographically tracking the judicial discourse around this incident. The inquiry commission, headed by a retired judge, will eventually issue an official report, which will join this event's other documentation in the state archive. 14 Usually, there is a large gap between a riot and the release of the state report about it. In this period, both judicial and nonjudicial discourse influence the course of the inquiry (Chatterji and Mehta 2007). Ethnography of archiving is precisely located in this gap.
A few works have already theoretically advocated the purpose of ethnography of archiving. Ann Laura Stoler (2008) maintains that the archive should be seen as a site not of knowledge retrieval but of knowledge production and of a potential ethnography of the state. She calls on historians and anthropologists to investigate what constitutes the archive, what form it takes and what critical features of state power their systems of classification and epistemology signal at and about specific times. Cohn (1987) has asserted that students of archival experience 'mine' the contents of government reports but rarely attend to their particular form or context. As Ranajit Guha (1987) rightly highlighted, 'sources' do not have real meaning in themselves and do not represent the truth. Whether documents are trustworthy is always a pertinent question, but attention to the social and political conditions that produced them would yield a more reflective account of the event under investigation (Bennett 2014;Carminati 2019;Trundle and Kaplonski 2011). Since the colonial era, the production of archives, especially commission reports, has been a powerful aid in ruling the population (Burton and Carlen 1967;Foucault 1977;Herbert 1960). This requires us to question the received categories in archives, and ethnography of archiving precisely serves the purpose of understanding the production of such categories. Commission reports, especially on riots, have always represented a delay tactic and a pause in policy, but they also involve complex power relations and the production of knowledge. An active ethnographic researcher at the site of document production can follow, in real time, the formation of 'social facts' and new social realities. The ethnographic site in this instance was primarily the Thiruvananthapuram courtroom, where actors involved in or with some other connection to the violence gave testimony during the commission inquiry. I was also present for the inquiry commission's activities in other investigative sites, especially Beemapalli.
The ethnography of an inquiry commission is also an anthropological account of a bureaucratic procedure that results in an official document. Emma Tarlo (2003), in her work on the state of emergency declared in India by Indira Gandhi, sees state officials as intermediaries and record rooms as field sites, both of which can provide enormous insight into the functioning of state power. Attendance at courtroom procedures and other activities puts researchers in the mediating space between the event of a riot and the archival report later made about it. Recent works on bureaucracy have shown that such documents are more than rational and formal emblems of state activity: they are are shaped by rules, ideologies, power, knowledge, practices and subjectivities (Feldman 2008;Gupta 2012;Hull 2012). It is thus important to study their creation, to grasp the ideological manipulations and power relations involved. This approach moves from what Hull (2012) calls as looking 'through' to looking 'at' these documents as constitutive and constructed.
After police lethally shot six Muslims in Beemapalli in 2009, it was pressure from civil society organizations and the community that forced the state to set up an inquiry commission, headed by a retired judge named Ramakrishnan. I began conducting my fieldwork soon after the violence occurred, and I followed the commission's court procedures and other activities. These revealed how narrative strategies and biased procedures were used to justify the state's violence against Muslims. I was present for the testimony of numerous state officials and bureaucrats, including police officers who were directly connected to the act of shooting. The inquiry, which spanned months, also invited victimized community members and other actors to provide their versions of the event and other evidence. However, the testimonies of state officials invariably dominated the discourse.
In court, I heard several police narratives, which paralleled the arguments seen in the archives on communal violence in colonial and postcolonial India. The reductive lens of the 'communal', which originates in colonial discourse, thus surfaced through official testimony reaffirming that communalism is an essential factor in any incident of this sort and religion is at the heart of the violence. One of the major questions of the proceedings was whether the police shooting which led to the Muslims' deaths was necessary at all. While the judge and lawyers of the victims raised this issue, most of the police officers justified the context and circumstance by invoking the notion that Muslims are 'communal', 'violent', 'criminals' and so on, painting the police action as reasonable and necessary. Balakrishna Pillai, then the deputy superintendent of police, stated in his testimony that 'the unfortunate incident at Beemapalli on May 17, 2009, was admittedly communal disturbances and the police action is quite justifiable as there was no other option to curb further communal clashes'. 15 C. P. Rajan, then a Circle Inspector sub-officer for Thiruvananthapuram City, described Beemapalli in very strong terms: Beemapalli Muslims are antisocial, fundamentalists and extremists. Muslims in Beemapalli live as a colony of their own and are frequently involved in communal clashes with neighbouring Christians. The geographical place itself has every appearance of an unruly locality. Police always find it difficult to negotiate with the people who live there; they have even beaten up police in the past. 16 Girish Kumar, another police official, likewise testified: It was a very critical circumstance, when police had no option other than firing. The police realized that Muslims were ready to riot, as they were carrying weapons and shouting slogans. If the police had waited for some more minutes, there would have been a riot, resulting in the killing of several Christians. Even the lives of police officers standing against the Muslims were in extreme danger. The police had taken every measure to pacify the violent crowd and nobody was willing to listen to them. 17 The first narrative shows that many of the orientalist views of Muslims seen in colonial reports are being reproduced in archives in the making, albeit with minor modifications in usage. The narratives of these state officials also mirror the usual argument in archival riot reports that state violence is justified to avoid the violence which the communities subjected to it would otherwise unleash. The strategy is to make the communities responsible for the violence against them and thereby prove the innocence of the police. Indeed, all the police officials ended their testimony with the statement 'Therefore the police firing was absolutely necessary'. A major component of this defense rests on the national and global discourse of 'Muslim terrorism'. On a day with intense arguments and counterarguments in court, one police officer cited an official report of India's central government to claim that the Muslims in the region have 'inter-state connections, and support of antinational forces operating from across the border'. 18 He then tried to establish Beemapalli's connection with many terror incidents throughout the country, thus painting the local Muslims as terrorists. These ethnographic instances reveal the symbolic violence involved in the making of an inquiry commission report, where violent stereotypes effectively justify actual violence by state actors.
I also directly observed how 'party formation' and other classificatory practices were constituted and operated in the courtroom. In the previous section, I emphasized the significance of understanding riot reports as discursive representations of divided subjects, and the courtroom experience gave a live picture of this process of division. In the Beemapalli case, there were three broad parties: the police and other government officials, the Muslims killed by the police and the Christians who fought with the Muslims before the police shootings. It was evident that the government lawyers and all the state officials supported this action by the police. Exploiting the past enmity between their communities, the police further strengthened their position by convincing Christians to testify against the Muslims. For Stanly Das, a member of the Cheriyathura Christian community, The police action is absolutely justifiable because there has been a long term communal mobilisation in Beemapalli. The problem of communalisations is only among Muslims, we do not indulge in such activities. Youth among Muslims play a key role in organised violent activity. They have connections with national and international organisations. It is clear that they have terrorist connections and there can be more harmful events in future, if the state officials and police do not act wisely in future. 19 Babu Samson, another Cheriyathura Christian, agreed: Beemapalli is known for illegal activities and violence. Communal force is so strong among them, as any slight altercation will provoke them to use physical force. Unlike in other places, violence is a normal thing [there] and they do not regret any activities they are involved in. Outsiders are always scared to visit their locality, due to this attitude. 20 I found that the Christians, like these two, who appeared for the judicial proceedings and those to whom I spoke during my fieldwork invariably used the same language about Muslims as the state. However, the courtroom ethnography also revealed how Muslims actively oppose such arguments, providing interesting insights into how contesting claims between the state and local communities characterize report making. For instance, Nizamuddin, a Muslim representative, had a very different view than the Christian witnesses: The police were biased in this whole incident. During the time of firing, they stood in the Christians' place and attacked us. When Muslims came together from our locality, the police assumed that we were going to attack them. In fact there was some kind of violent situation, but the situation was not that bad to fire at people. Christians used the police to attack us. 21 Abdul Azeez, then the president of the Juma-ath, echoed this view: The police have done a lot of injustice to us. They have a lot of prejudice. When our people were involved in some issues in the past, we always helped the police to arrest them. The police always proclaim that there is communal tension in our locality despite the harmonious relationship we maintain with the neighbouring Christians. They need to portray the incident as communal because they want to justify the firing on and killing off our innocent people. Is there any example of any injury to Christians or any injury to policemen to [justify] kill[ing] our six people. 22 These narratives are reminders that even overwhelmingly statist riot reports will include voices of dissent. Finding them, as I mentioned in the previous section, requires conscious effort and creative dialogue with the narratives. Here, in the 'real' field, the circumstances of the construction of 'voices' were clearly visible, and 'listening' and engaging in 'dialogue' were everyday processes that allowed the interrogation and disruption of 'established' truths in archival materials. As for the state itself, its perception of the event is indicated by its classificatory practices at the site of Beemapalli. A 'line of control' separates the Muslim and Christian neighbourhoods, just like the territorial boundaries between unfriendly nation-states, such as India and Pakistan. This 'line of control' and the everyday police surveillance of the area give it the appearance of a trouble spot even when things are perfectly calm.
An instance during my fieldwork revealed a problem in the state's collection of evidence for the riot report. When the head of the inquiry commission, Judge Ramakrishnan, and his team visited the area, I was present, along with a large crowd of locals, to watch how the commission worked. The officials divided into square meters the locations where the police were said to have fired. The exact location was unknown, and the locals offered contesting claims. Before the officials arrived at their conclusion, Ramakrishnan asked the crowd about where these shootings took place and about the properties that were destroyed on the day of the incident. The villagers, who were emotionally charged, started giving contradictory answers, often shouting at the officials. For example, when Ramakrishnan asked where Firoz (the youngest victim) was shot and died, a few said it was on the football pitch, others claimed it was the cricket ground. It was clear that the questions were meant to manipulate the case and bolster the statist view. I got the impression that the commission was happy to receive these emotional and contradictory responses, which it could use, following the path of so many previous riot reports, in a key narrative strategy to dismiss this community's opinions as illogical, in opposition to its own rational observations and conclusions. This ethnographic insight, in turn, helps to question the claims of 'rational' investigation in older archival reports and to dissect their complex narratives for a more effective archival ethnography.
Ethnography also shows the deeper implications of report making on the communities affected by the violence. At the location, the commission's visit can provoke rumours and even disruption. In court, testimony recalls not only the incident under investigation but also earlier events and wounds. For many individuals and perhaps the whole community, it may be impossible to move on until the judicial process has finished. Even those who do not attend court will hear public announcements by mosques or churches meant to encourage their members to participate in the inquiry. All of this keeps the past event present and hinders the fading of enmity.
The insights gleaned from ethnography of archiving reveal how the making of a riot report is characterized by ideological manipulations, power relations and symbolic violence. The nature of the interactions and procedures of the judicial inquiry process make it far from the neutral and rational enterprise of a responsible government. Indeed, the everyday, interactive field of bureaucratic performance gives rise to a series of problems in the archival reports on riots even in postcolonial India, and judicial discourse can marginalize victims' pursuit of justice and provoke hatred between communities as much as the horrific memories of the violence under investigation.

Conclusion
This paper has aimed to extend the debate on the synthesis of archive and ethnography to 'communal' violence and riot studies in India, offering the example of a nuanced methodological intervention for a better understanding of riots and of riot inquiry commission reports in India. It first reflected upon the problems in archival representations of violence, such as colonial genealogy and the pervasive effect of generalized ideology on government reports in strikingly specific contexts, such as Christian-Muslim violence in coastal south India. Further, by elaborating on key concerns in the debate over ethnography and archives between anthropologists and historians, the paper demonstrated how 'archival ethnography' can enable a nuanced reading of riots and riot reports in postcolonial India: despite all the problematic assumptions in archival sources, they are a potential ethnographic site where the subjects' voices can be heard, allowing the researcher to engage in dialogue with their contents through rereading and reinterpretation. Finally, this paper proposed a nuanced methodology called 'ethnography of archiving', which ethnographically tracks the making of an archive or a government document on a riot (or series of riots). This endeavour, by following the judicial and non-judicial discourses, bureaucratic procedures and other official activities that inform the creation of a government report on violence, offers the possibility of unravelling the power relations, mediating processes, manipulative moments and bureaucratic performances that make commission reports problematic even today. Notes 1. I distance myself from the word 'communal' attached to violence, at least in characterizing the cases discussed here, for reasons I elaborate later in this paper. However, I use the term throughout, in quotation marks, because it is the predominant trope in almost all studies of violence between members of different religious communities in India. Likewise, I am aware of the differences between violence, riots, pogroms, etc. However, the inquiry reports I investigate here employ all of these terms, and I sometime follow these uses, for the practical purpose of my analysis. 2. Between these two extremities, political scientists and sociologists, for instance, employ methods that include both fieldwork and archival research. 3. For instance, anthropologists who have conducted fieldwork on riots have also consulted archival reports on riots to analyse them. Hansen's (1999Hansen's ( , 2001 works are good examples. 4. This paper uses 'archival records', 'government reports', 'commission reports' and 'riot reports' interchangeably. 5. Violence among religious groups has been approached from several perspectives, such as Marxist, instrumentalist, psychoanalytical and postmodernist. Though they disagree on many counts, this assumption is common to all. 6. Poovar is a village in Thiruvananthapuram near Kollencode. 7. District Magistrate, Trivandrum, to the Chief Secretary to the Government, November 25, 1939, ROC No. 1382, Travancore Archives, Central Archives, Thiruvananthapuram. 8. The Arattu is a holy bath taken as part of a ritual at major temples, usually at the end of major festivals.