Riots, masculinity, and the desire for passions: North India 1917–1946

ABSTRACT This article aims at a threefold historicization of riots, drawing on sources depicting communal riots in North India in the twentieth century. (1) The concept of ‘crowd’ covers a large variety of phenomena. Rioting crowds could range from several tens of thousands, gathered at one place, to just a handful of men performing the actual arson and murder in the back-lanes. (2) Not all rioting crowds give rise to the same emotions. Research on riots often focus on anger and fear, but the sources frequently speak more about josh, about excitement, enthusiasm and fervour – emotions which participants in a riot felt they needed not only to experience, but to show, and proudly show. (3) Finally, the violence itself underwent decisive changes between the 1900s and the 1940s, not only it whom it addressed, but also in the weapons used and the forms of pain it aimed at inflicting. Taken together, the historicization of the crowd, emotions, and violence allows us to overcome a number of dichotomies: Between the riot that essentially always remains the same, and the riot that can only be viewed as a fragment; between the learned and the spontaneous; between emotions and interests, not only by showing that interests generate emotions, but also by showing that the actors had a vital desire to experience certain emotions, which would transform them and thus change the fate of their community. This does not depoliticize the research on riots, but it brings in politics in a more complex way than reducing deadly emotions to the manipulation of the elites.

Riots in South Asia have been intensively studied in the past decades. We have seen debates on their historical origins, whether they were developments that had come to the fore already in the seventeenth and eighteenth century, or a result of British colonialism; on their moving forces, whether elites manipulated the crowd, or subalterns pursued their own agenda; on the role of interests and emotions. Individual riots and regional or temporal clusters of riots have been explained through their embedding in social, economic, and political developments; the results of riots, not least their role in the partition of the subcontinent, have been investigated in detail. More recently, the focus has shifted to the experience of the victims, to trauma and to memory studies. 1 While we know much about what happened before or after the riot and about its causes and effects, this article focuses on the riots themselves. Taking its cue from earlier historical investigations and from methodological reflections in anthropology and political science, 2 it attempts a threefold historicization: of the crowd, the emotions, and the violence performed.
Crowd: Crowd psychology suggests an image of large numbers of people, densely packed, moving together as if guided by a single will. Affect studies, though guided by different questions and, as a rule, quite critical of crowd psychology, meet this approach in their emphasis on the

Riots in North India
This article looks at riots in Delhi, the United Provinces, and Bihar between 1917 and 1946. Existing scholarship often tends (with good reasons) to distinguish between communal riots, mass actions linked to the national movement, and agrarian or labour movements, focusing on either one of those to the exclusion of the others. For the questions centred on the event of the riot, which the present article attempts to address, this distinction is less important: Even where the actors came from different groups -and often the lines between them were blurred -they drew from a common emotional vocabulary and from the same repertoire of political actions, violent and non-violent. A riot could well start as a nationalist or labour event and then turn communal, or be limited to one community, drawing on strong religious emotions, but confront the colonial power.
I am drawing on two types of sources: The colonial archive, mainly the reports written by the district magistrates and superintendents of police, the inquiries after the bigger riots, and the court proceedings; and the Times of India, which has not only been digitized, but permits a full-text search. 7 Together they cover around 200 violent incidents. While I'm not aiming at a statistical analysis, this broad basis ensures that the detailed readings are indeed representative of larger trends. I include the colonial selection and translation offered by the Newspaper Reports of the United Provinces. An in-depth analysis of the original vernacular newspapers, however, has to wait for some other occasion. 8 Four riots have been chosen for close reading. The Shahabad riots in Bihar in 1917 originated in the cow protection movement, which had taken roots in the region since the last decades of the nineteenth century, and had already, at several instances, led to local violence. What is remarkable about the series of incidents in 1917 is their extended character, involving most of the region between Arrah and Aurangabad, on both sides of the Sone river. The crowds attacking the villages were huge, the numbers given vary between 20,000 and 50,000 participants. The riots started after a compromise concerning the sacrifice of kine at Bakr Id in Ibrahimpur -Muslims would forgo the slaughter, and in exchange, the Hindus would renounce baiting pigs near the mosque during the Gaidar festival -had broken down. However, the extent of the region involved in the violence, which offered few possibilities of transporting huge crowds except by walking, meant that the mobilization could not be explained as a spontaneous reaction, but had been organized well ahead. 9 The next riot, taking place in Delhi in 1924, can be read as an example of the deterioration of communal relations after the Non-Cooperation movement had been called off in 1922, and the alliances which had sustained it broke apart or had to be reconfigured. Delhi had become a centre for the organization of Muslim Tabligh and the Hindu Shuddhi movement, aiming at the consolidation of these two communities and the conversion and reconversion of those deemed to have left the community to which they belonged. In this tense atmosphere, relatively small incidents sufficed to spark off violence. In July 1924, several of these disputes came together: a property dispute at Bara Hindu Rao, involving a disused mosque, a quarrel between a Hindu boy and some Muslim porters, and finally, as Bakr Id was approaching, the question which routes would be assigned for driving the cows from the houses, in which they were kept in the final weeks before the festival, to the slaughterhouse outside of the city. 10 The Kanpur riot of 1931 is probably the best studied riot in the interwar period, not least because it has produced ample documentation and two reports, of several hundred pages each, by the government and by the Congress. As in Delhi, the conversion movements and especially the organization of their volunteer associations played an important role in creating an atmosphere of distrust between the communities and an expectancy of a future battle. Kanpur had seen no less than four major riots in the 1920s; two labour disputes turned violent and two communal riots. The Civil Disobedience movement did not renew the scenes of communal amity which had marked the earlier Non-Cooperation movement. Most Muslims stood apart and did not join the boycotts and hartals, the closing of businesses as a sign of protest against a moral injustice. It was such a hartal that triggered the riots in March 1931: While the initial protests against the execution of the revolutionary Bhagat Singh were directed against the State and against Europeans, the refusal of the Muslims to close their shops turned the riot communal and led to mass violence engulfing the entire city and its suburbs. With more than 120 killed and several hundred injured according to conservative estimates, it was the deadliest riot in North India in the interwar period. Throughout the violence, the police remained mostly passive, either under instruction from the magistrate or due to lack of instructions. 11 With the final riot we return to Bihar at the beginning of the Partition riots. After the Muslim League had called for a Direct Action Day at Calcutta on 16 August 1946, the city had been engulfed in violence at an unprecedented level for several days. Through the networks of Muslim labour migrants in the city, rumours had spread to Noakhali in eastern Bengal, leading to the killing of thousands of local Hindus. Bihar followed suit within a few days, again with rumours of violence, forced conversion, and abduction fanning the flames of vengeance and counter-vengeance. Within a week, more than 5,000 people lost their lives and 70,000 refugees had to be accommodated in camps. The questions to be settled through riots no longer pertained to individual villages or towns -what was at stake was the question whether Partition had become inevitable (or could be made inevitable through violence), whether it would also involve the partition of Bengal, and whether parts of Bihar would belong to Pakistan, in whatever geographical configuration Partition would take. 12

The passionate crowd
The concept of the crowd evoked by canonical works such as Gustave Le Bon's work on crowd psychology relies on the image of a large number of people and assumes that 'from the mere fact of their being assembled, there result certain new psychological characteristics.' 13 Among these characteristics, the emotionality of the crowd stands out for Le Bon. These emotions are neither willed nor learned, but imbibed by everyone through a contagion based on bodily proximity. In turn, these powerful emotions are the reason that rationality, individuality, and consciousness are lost and give way to a collective unconscious, which can either be manipulated by clever leaders, taking advantage of the crowd's suggestibility, or be seen as the expression of the collective's racial memory -here Le Bon remains equivocal. 14 Le Bon has been strongly criticized for his elitism and racism. 15 Nevertheless, some of his ideas still inform present day ideas about crowd behaviour: The fusion of individual selves; the hydraulic model of pent up emotions, which erupt and after a certain time spend their energy and calm down; the idea of emotional intensity and suggestibility 16 ; and, more recently in the guise of affect studies, the idea of bodily contagion. 17 Though the British colonial officers drew some of their interpretation of collective violence from a vulgarized Le Bonian psychology, this type of crowd does not often appear in the descriptions the sources provide. An exception, however, might be the Shahabad riots of 1917, during which the crowds attacking village after village numbered several thousand 18 and 'were so densely packed that they had to hold their lathis over their heads.' 19 On the other hand, bringing these crowds together from a wide stretch of territory and for several days required careful planning, including local hospitality. 20 While it can be argued that emotions gained in intensity during the actual performance of violence, the decision to aggressively prevent the sacrifice of cows was taken before the crowd physically congregated as a crowd, whether this decision pertained only to one specific village or to the entire district, and whether its aims were restricted to a single year or conceived as 'a general movement to so intimidate the Muslims that they would never dare to sacrifice a cow again.' 21 The image of a large and anonymous crowd is more difficult to fit to urban riots. Some of the larger thoroughfares in Delhi and Kanpur may have accommodated several hundred or even several thousand persons. However, it was the narrow by-lanes of the old city that bore the brunt of the violence. Spatial restrictions meant that the crowds that were active here had to be much smalleras a rule, groups of fifteen to twenty young men were at the centre of the looting and sacking of shops and houses 22 ; while what was referred to as 'stray assaults' in the language of the police and the newspaper reports required even fewer persons, often not more than four or five. This argument does not imply that there were no violent encounters between larger groups -often the riot started with a dispute over a procession or with a clash between two processions on the main thoroughfarebut this does not seem to have been the most frequent, the most sustained, or the most deadly form of violence.
Crowds, whether small or large, also rarely stayed together throughout the riot. Contagion triggered off by tightly packed bodies might have happened at one point, but closer attention to the micro-temporalities of the incidents shows that they do not proceed along a single arc, from an explosion at the beginning to a slow fading off at the end, once the energy had been spent. Rather, the groups that make up the crowd assembled, dispersed, and reassembled over several days. Phases of intense violence gave way to times of negotiations, or simply to lulls, during which the rioters returned to their families, perhaps had a meal and some rest, only to go out again later at night or the next morning and resume once again the same violent intensity of feeling. 23 This does not negate the correlation between strong emotions and acting in a crowd, but it calls into question assumptions about the quasi-natural way in which passions are aroused in collective action. It also moves beyond the alternative of seeing passions either as a smoke screen hiding the 'true' motivation or as depriving the actors of their agency. The excitement that the young men experienced during crowd violence was not (only) inherent to the nature of crowds, but had a lot to do with the way they have learned to perceive and to value passions and the actions and practices to which those led. Acting in a crowd and as a crowd, I suggest, was a powerful tool for producing the desired emotions of heroism and martyrdom, but also of youthfulness and virility, which were emphasized again and again in pamphlets as central to the regeneration and survival of the community. We will come back to these emotions in more detail in section 4.
But not everyone, of course, rejoiced in the riot -some joined the fighting because in this situation it seemed the only possible way to protect their family and survive. Munshi Nur Ilahi, for example, reported how he had gone home to his house in Kanpur with the intention of stopping the fighting in the neighbourhood, but then found himself 'rather inclined not to interfere with Muslims, as, if I had stopped them, the Hindus would have rushed in, and so to say our whole family's lives and houses were in danger.' Instead, he started throwing bricks from the roof of his house, together with his nephews. 24 Some were unsettled by the scenes they had witnessed, while many more tried to distance themselves from the violence and even to actively prevent it.
Those who did not subscribe to the rejuvenating power of josh and refused to legitimate the violence they witnessed (or at least some of it; positions were rarely black-and-white in this context), still needed to explain the events. This became especially urgent if they opposed the traditional colonial narrative of the Oriental either being naturally prone to violence or stuck in an earlier stage of development in which he had not yet mastered the control of his passions. One possibility was to establish an internal differentiation between those who belonged and outsiders, whose passions and lack of control did not require any explanation precisely because they were outsiders. These could be migrants, who lacked roots and stability, and people arriving from elsewhere with the explicit aim of rioting, 25 but also 'inner outsiders,' badmashes and goondas, translated into Anglo-Indian parlance as 'bad characters.' In the interpretation of the Kanpur riots, this category included drug dealers and groups of strongmen, at times hired by businessmen and politicians for protection and to further their interests. 26 But depending on the perspective, the lines between criminals, volunteer organizations of different associations, and labour were blurred, potentially including anyone whose social position or activities lacked respectability in the view of the speaker. 27 If this inner differentiation was not enough and it was 'the people' who took part in the riots, their emotions could still be interpreted as not representative of their real identity, but the result of manipulation, be it by agitators (this was generally the version of the colonial state) or by the British, who had fomented communal passions in an attempt to divide the country in order to perpetuate their rule. 28 In this picture, emotions were no longer something that could be chosen or controlled, but a simple reaction to outward stimulants, which made them easy to provoke and direct to any chosen goal. However, the discourse was not homogeneous in this respect, even within the same text. The simultaneous exhortations to resist manipulation, control the passions, and opt for a nonviolent struggle were based on a different interpretation of emotions, which foregrounded moral considerations -to strongly feel the virtuous emotions and refuse to give in to all others -and implied that people had a choice.

Communication before, during, and after the riot
One result of the traditional division between emotions and rational debate, at the basis of much of the theorization around the development of the public sphere, is that it forces us to decide between either focusing on the violence, its causes, forms, and results, while neglecting the negotiations accompanying almost every riot, or bringing in the negotiations and reducing emotions to an epiphenomenon. In this interpretation, passions that still allow for consultations and compromise cannot be 'real,' and people involved in these negotiations are themselves rational and thus devoid of emotions, while at the same time manipulating the crowd, turning their emotions off and on at will. This section argues that we can gain a more complex picture by tracing in some detail the moments in which specific forms of communications are possible and when they become difficult or impossible, and how communication and emotions are not linked in a zero-sum game, but interpenetrate and shape each other. 29 Most riots, at least in those instances when conflicts could be anticipated, were preceded by negotiations between the parties. This is even truer for the overwhelming majority of cases in which a clash of interests did not lead to violence. We would do well to remember that for every Bakr Id riot in a village or neighbourhood, there were hundreds of villages and neighbourhoods where no clashes occurred, where the routes of processions were designed and respected in order to avoid conflict, and where neighbours continued to live together peacefully in spite of potential conflicts. Communication aiming at a compromise took place, very often successfully. Frequently, but by no means always, these conversations were initiated by the colonial officers, who were later tasked with backing up the agreement by force, if need be. In the case of Shahabad, which we have briefly considered above, a compromise had been reached in a number of villages. Muslims forwent the slaughter of cows on Bakr Id in exchange either for financial help (goats were more expensive then cows) or for the Hindus in turn accepting restrictions on one of their own festivals, which involved the baiting of pigs near the local mosque. 30 In Delhi, the routes assigned for driving the cows to the slaughterhouse had been fixed for many years; however, wishes to have them altered were discussed regularly. In 1924 (and again in 1926 and 1927), it was the question which of the streets behind the Fatehpuri Mosque qualified as 'Muslim' or 'Hindu' enough to make the sight of cows designated for the Bakr Id a right or an offence. 31 These strategies for conflict resolution depended on two factors: First, the parameters according to which decisions would be backed up by the British needed to be clear. This became problematic by the last two decades of the nineteenth century, when the policy of not disturbing existing traditions and agreements clashed with a new attempt to do justice to those whose religious observances had been curtailed earlier (or those who made this claim). This widely opened the space for negotiations and led to a struggle to finally settle scores and establish the community in its rightful position. Second, the leaders entrusted with the negotiations had to be able to wield enough influence to ensure that the compromise would be accepted and adhered to. This too could no longer be taken for granted, especially since the First World War. The Non-Cooperation campaign and the reform movements in all of the communities brought new leaders to the front, who often challenged the authority of the traditional leaders and their right to speak for the community and to represent it. 32 Parallel to these negotiations, mobilization for a potential conflict was already taking place. In so far as emotions played a role in this mobilization, here they worked outside of the pressure of a crowd, though they were rarely individualized. The report on the Shahabad riots contains a number of patias, letters distributed through a snowball system, which were circulated in the days preceding the violence. They exhort their readers/listeners to save the cows, to snatch them from the Muslims, and to be careful to take the right side in this fight. Instead of helping Muslims and hiding them, readers were exhorted to loot Muhammadan houses and kill their residents. In this they would have the support not only of the local Maharaja, who would immediately send 'a force armed with guns' if required, but also of Hindus from all over North India, and even of the German emperor. If, on the other hand, they refused to do as they were told, this would be counted as equivalent to slaughtering a cow, 'having sexual intercourse with your daughter, drinking the urine of your wife and cohabiting with your sister's daughters.' In such a case, it would 'better that you should marry your mother with Muhammadans.' 33 This can be read as an attempt to shame the addressees into action by questioning their sexual integrity and thereby their position as guardians of their families -a direct linkage between cow protection, male honour, and patriarchy. At the same time, if these equivalences would be backed up by the local community, they would constitute the kind of 'sins' traditionally leading to excommunication. Thus, taking part in the riots and being seen as doing so by the village may also have been a 'rational,' interest-based decision. 34 In turn, this need not efface the emotional load: Shame and the fear of the social death of excommunication might not have been less intense feelings than love for the cow or the hatred for its slaughterers.
The case of Kanpur is slightly different, as the mobilization did not pertain to one specific goal, but was directed towards what was increasingly perceived as an almost apocalyptic fight for the future, which would decide the community's life-chances, honour, and even survival. 35 Rivalling movements sought to increase the internal coherence and the fighting power of either Hindus (sangathan) or Muslims (tanzim) and prepared them for the impending battle. In accordance with the promotion of a youthful and martial spirit, of 'young men who are ready to sacrifice their lives,' 36 these organizations were mostly composed of young men, often but not necessarily of low social status, and linked to the traditional akharas, wrestling clubs with a guild-like structure. 37 Since the late 1920s the tanzim movement had shown an increasing presence in Kanpur, as in many other North Indian towns, partly in response to the increasing local activity of the Congress, which was not always clearly distinguished from the promoters of Hindu movements, such as the Arya Samaj. The tanzim movement and the Arya Samaj took out processions whose participants were uniformed, carried flags, and displayed arms, mainly swords and spears. 38 Both sides performed martial songs, in which they threatened each other. If the Muslims threatened, 'we will annihilate the life of the unbelievers' (ham hastiye kafir ko mita denge), 39 the Hindus replied, 'we will annihilate the Muslims from this world' (ham hastiye Musalman ko dunya se mita denge). 40 If the Arya Samaj threatened 'to raise the flag of Om over the Kaaba' (Om ke jhande ko Kabe men laga denge), 41 the tanzim procession sang how 'the temple would crumble under the blows of Islam' (Islam ki thokar se mandir ko gira denge). 42 Other songs were much more elaborate, calling the Muslims to awaken from their slumber and to remember what they had done while they had josh for the unity of God (jab josh men wahdat ke ayenge), and when the call for jihad had joyfully awakened their passion and their willingness to sacrifice everything on the path of God (Agar mauqa-e jihad aye khushi se josh men akar/Katarkar sar Khuda ki rah men sarkar ban jao), enabling them to destroy all traces of the enemy. 43 Both sides revelled in visions of violence leading to future glory. Violent emotions did not emerge at the moment of the riot, or at least not without this prehistory. Rather, they were deeply embedded in a worldview and in an understanding of temporalities -the question for them was not if, but when the battle would take place.
Unlike what the image of the crowd overwhelmed by emotions suggests, negotiations did not stop once the riots began. The reports contain a large number of narratives of leaders trying to stop the violence. This could take the shape of calling and moderating meetings between leaders of communities at a very local level, as the compromise negotiated effected between the Jats and the butchers of Sadr Bazaar in Delhi in 1924, which led to the reopening of the shops and thus marked the return of normality, 44 or the meetings which Dr. Ansari convened in Hakim Ajmal Khan's house to end the riot in Delhi three years later. 45 However, it could also mean that leaders walked right into the spaces where violence was taking place and tried to talk to the rioters -often successfully, but also at a high personal risk. The Kanpur Congress leader and editor of the Hindi newspaper Pratab, Ganesh Shankar Vidyarthi, addressed the people from the first day on, reminded them of Gandhian non-violence, and scolded them for shouting 'Mahatma Gandhi ki jai' while looting and killing. His authority cut across the boundaries of communities. His influence rested on his personal moral authority, but also on the Gandhian emotions he embodied and publicly displayed, on his empathy with the victims and his tearful prayers at the sight of a slain child. He was able to stop the violence in a significant number of cases, before he himself was murdered by a group of Muslims resenting this interference. 46 The most prevailing form of communication before and during riots was the rumour, information on what was happening (or had happened) at other places, either during riots immediately preceding the actual one or simultaneously in other neighbourhoods or adjoining villages. These reports of violence often triggered new violence, or if they were fictitious, initiated it in the first instance. 47 The state -the colonial powers, but also the Indians in the municipal boards and, after 1937, in the ministries -tried to check the rumours through regular public announcements and bulletins, but also through warnings to the press and to those who could be shown to be actively spreading rumours. 48 These rumours are amongst the most difficult sources to analyses. They certainly do not offer any 'true' picture of what happened. However, they also cannot just be discarded, as their gory details and their elaborate descriptions turned them into powerful influences on the imagination of what a riot looked like -not just this specific riot, but all riots. This allowed those whose task it was to report on the violence to fill the blanks in the sparse information they had: Already before the riot took place, everyone assumed that they knew what it would look like. 49 This might be one of the reasons that the newspaper reports on the riots read so repetitive, and the more so as the riots became more frequent. The journalists already had the plot at hand and only needed to add some local details, the name of the town, the festival or procession, whether the police had fired or not, and the number of dead and injured. This repetitive knowledge of what happened during a riot, however, also worked on another level, providing the perpetrators of violence with a pre-mediation of what they would see during the riot and how they would feel. For hardly anyone were the crowd and the riot an entirely unmediated experience, something that unexpectedly overwhelmed them on a bodily level and made them feel passions they had never anticipated. This pre-mediation offered not only a grid for interpretation and the ascription of meaning, but also created a lexicon of available forms of violence. While older layers -possibly pre-colonial, but also the memories and narratives of 1857 -remained present in this lexicon and could be reactivated and reinterpreted, new forms also developed. 50 What made the generic descriptions of the riots so dangerous was not just the provocation to vengeance, but also the way they endowed violence with norms and normality -even if the rumour was not true the first time, it offered a blueprint for future (re-)enactments, over and over again.
Finally, communication was crucial for the termination of the riot. Even once the violence had been quelled, a riot was not over before the shops reopened, and before the processions had properly come to an end, the tazias buried, the Holi burnt, and the idols submerged in the river. Especially the closing and opening of shops, hovering between the hartal -a protest against a moral injustice -and a precautionary measure to prevent looting, marked the boundaries between ordinary and extraordinary times. Discussions with the shop-keepers, therefore, not only involved questions of trust, but were central in practically and symbolically ending the riot: If closed shops were the reminder of the persisting moral injustice, their opening stood for the resolution of the conflict. 51

Hailing the state?
What was widely shared across different forms of political mobilization was a perception of temporalities: The future was no longer given, but something humans had the possibility and even the duty to influence. What it would bring for them and for their children was not preordained by fate, but was in their own hands. Even God would only help those who helped themselves. The time available for shaping this future was increasingly short. Swaraj might happen next year, as Gandhi announced, but even for those who thought this prediction too optimistic, the end of British rule was something they expected to take place in their own life-time. This gave an importance and urgency to the present, which was new. Finally, the creation of the future was going to happen through battle. This battle might take different shapes in the mind of different people, but even for Gandhi, it was clear that a non-violent battle was still a battle and the satyagrahis were soldiers, if of a particular sort. 52 This was a far cry from the emotions of civility and balance advocated in the nineteenth century, though of course, questions of gender and age mattered. 53 While the desire for passions was least contested for young men, it also affected many middle-aged men and even women. Though many men remained uncomfortable with the idea of passionate women, from the mid-twenties onward we find newspaper articles exhorting women to carry poison and commit suicide rather than become the means through which their community could be dishonoured. 54 What became dominant was a cluster of visions of masculinity, which shared a family resemblance across boundaries. It influenced the volunteer organizations of the religious communities, but also of the Congress, and marked at least part of the labour movement. The masculine body was a body which had to be trained for battle, violent or non-violent, a body which had to be disciplined through exercise, through uniforms, and through the bearing of flags and other insignia. 55 This training and selftraining was needed to produce the proper body, but also the emotions appropriate for a fighter. Bodies and emotions were closely knit together in the perception of the actors -new bodily ideals could not be conceived without a desire for the corresponding passions.
The need to overcome cowardice was common to all groups. For Gandhi, cowardice was the greatest sin; so great, that he preferred violence to a lack of courage. 56 The fighters -the volunteers, but also everyone who would be joining the battle for the future -had to burn with josh, with love for the nation, or the community, or the cow, or God. The object of this passion and its precise content -love, devotion, zeal, fervour -almost disappeared behind its intensity, behind the need and the yearning to strongly feel the emotion with boiling blood 57 and 'red eyes, one stare of which once consumed the earth, [and which] are full of those sparks.' 58 The metaphor of fire, which is central to so many texts of the time, is fascinating in its ambivalence: Fire is related to love, but also to anger. It is the purifying passion, burning away sins and whatever stands in the way of renewal; as such, it is also linked with spiritual power. But fire is also associated with danger. Like the fire of a volcano, riots erupt with the destructive energy of a natural catastrophe, which cannot be halted until it has exhausted itself 59 : 'Swallow the blazing fire and its sparks and the red streams of the glancing and shining swords. Burn the revolving sky and reduce it to an abject plight. O negligent one, the fun is if you also shed blood in revenge for blood.' 60 This josh linked to the fire of love and anger was deemed central for the rejuvenation of the community. Not only in India, the confidence in the civilization of mankind had given way to new anxieties that an excess of civilization might sap the vitality of a community, producing timid and weak men (they were less worried about women). These men would move tamely, always in accordance with the rules society and the state imposed on them. They lacked the most essential trait of character: the resolve and manliness to take up responsibility. Interestingly, a good number of articles read the exhortation to non-violence within the framework of the older discourses on civility and emotional restraint and worried that it would sap the strength of the nation: 'The sermon of ahimsa has emasculated the Hindu nation.' 61 What was needed, many claimed, was a new generation, in touch with the power of their youth and their virility, fearless and ready for sacrifice and martyrdom. 62 The implications of these ideals have not been fully worked out yet: What did it mean for the social structure if it were the lower classes who could display stronger bodies and more untamed passions? 63 How did families deal with the new leadership role given to the young men?
Long before an individual riot started and the crowd came together, the experience of passions had thus already become an ideal to aspire to. These passions could be generated by violence, but for others they could also be produced by coming together in a crowd of satyagrahis and facing police brutality without flinching. Passions endowed actions with a legitimacy which went far beyond anything that arguments could bestow. If the most youthful and passionate peoples would be the only ones surviving the future battle for existence, feeling passionately had to become the linchpin on which the fate of the nation depended.
However, not everyone shared this move from balance to strong emotions as the new social ideal. Hakim Ajmal Khan, who had been active in promoting cooperation between different groups in Delhi for many years, ascribed the riots of 1924 to the inflammation of the 'worst passions' by the public press, which had generated a 'general atmosphere of mutual distrust and suspicion.' Instead of giving in to the 'force of mere passions,' the general public needed to return to the 'mental equilibrium' that they had lost. 64 Against the current, which claimed salvation through josh, here it was the return to the traditional Aristotelian concept of etidal, of balance and justice, which was advocated. 65 Passions needed not only be felt, but also expressed and shown, in order to fulfil their role. Much has been written in recent years about the development of 'hurt sentiments' as a legal category. 66 While colonial law had been created in an attempt to provide a procedure to peacefully resolve issues of hurt religious sentiments -after the trial of the author of a scurrilous pamphlet insulting the Prophet had resulted in his acquittal, the eruption of riots, and his subsequent murder -the results were more complicated than anticipated. Linking the issue of protection by the state to the existence and public display of hurt sentiments created an incentive for the performance of outrage, in a situation in which, as shown above, violent passions were already increasingly culturally validated. What seems to be taken for granted in these analyses is that it was the state that was the first and most important addressee of this display of emotions -what Lisa Mitchell has so aptly called 'hailing the state.' She points out that riots and public demonstrations did not rupture communication with the state or ignored it in favour of counter-publics, but on the contrary, constituted a different form of engaging it by calling its attention to injustices and suffering. 67 This can be shown for many of the riots since the early twentieth century, most expressively in Kanpur 1913, when the Muslim Gazette from Lucknow exclaimed: 'The Cawnpore mosque incident has inflicted such a deep wound on the hearts of Muhammadans all over India it will never be healed and will continue to discharge pus until the Day of Judgement. If it had been possible to tear open our breast and to show that wound we would surely have shown it to His Honour [the Lieut. Governor of the United Provinces].' 68 The state was not only the power which could afford an effective remedy to hurt sentiments -before the enactment of Section 295A, and even more so afterwards -it was also identified with the British discourse on the Indians' lack of civility, on their weakness and cowardice. This made it an ideal addressee for demonstrations that these accusations were not only false but, on the contrary, the demonstrators were endowed with a fine-tuned sensitivity and did not fear fighting and -if need be -dying for their convictions. If Hindu newspapers exhorted their readers, 'We wish that every Hindu should today get rid of cowardice and slavishness and be immersed in the spirit of communalism and nationalism,' 69 their Muslim counterparts responded, 'There are occasions in the lives of nations and people when, to put a stop to an unchecked process of dissolution and decay, courting death is considered better and more honourable then clinging to a shameful life -life of unsanctioned and unholy bondage and slavery.' 70 However, in the years after the War, displaying wounds, showing passions and the desire for martyrdom was increasingly directed not only at the State, but also at the communal other. The different communities and constituencies were very much aware of what the other was doing and saying, and it is to this that they primarily reacted. This ranged from copying organizational patterns like volunteer associations, the wearing of uniforms, and training through akharas, to taking up each other's songs and just replacing the name of one community with that of the other. But it also involved developing a repertoire of emotional expressions and languages for public display, which, notwithstanding their differences, responded to each other and could be understood by the other.
This does not necessarily contradict Markus Daechsel's argument that Muslim politics in the interwar period was increasingly directed towards self-expression at the cost of the communication central to both politics and society. 71 However, this self-identification as the right kind of person, as a moral subject, virile and with strong passions, was developed in response to other groups' self-identifications. It was not turned inside so much as proclaimed publicly, through speeches, through the media, through demonstrations, and through violence -all of which might be read not so much as the opposite of communication, but as its very essence. Hailing the world, hailing the state, hailing the communal other, and hailing one's own peers and convincing oneself are not be neatly separated, but form parts of the same process.

The language of violence
Riots, the preceding two sections have shown, are not a disruption of communication, but its continuation in a different form: 'Acts of violence have acquired their own semiotic register, a singular form of political "communication."' 72 Though the linguistic register and the register of collective violence are profoundly intertwined, they cannot be completely mapped onto each other -if the linguistic register has a history of its own, so has violence. Forms of violence change over time, even if at first sight the newspaper reports might seem repetitive in their descriptions.
The sociologist Jan Philipp Reemtsma has proposed to divide violence into three categories. Locative violence is interested in the body of the adversary only in relation to its position in space; it aims at moving bodies from or towards a certain space. What matters is the space, not the bodies as such. 73 Raptive violence aims at power over the body, in most cases in order to use it sexually; autotelic violence aims at the destruction of the body for its own sake, often using forms which are perceived as especially humiliating, so to annihilate not only the body, but the subjectivity of the victim as well. 74 For our context, I would suggest taking the latter two categories together, as the violence against women in most cases was not addressing the women, but the men who claimed responsibility for them.
My argument is that we can trace a development from locative to autotelic violence in North India in the nineteenth and twentieth centuries. This does not mean that the earlier forms completely disappeared. It also does not mean that earlier forms of riots were not cruel or that rapes did not occur, but rather that neither the cruelties nor the rapes were the primary feature of the violence performed. The prototypical case for locative violence in North India would be the dispute over procession routes. The ability to claim the use of certain important streets and places for a celebration was a way to adjust the changing claims of different communities to prestige and power. If this result could be obtained through negotiations, then violence and, more so, excessive violence, was not only unnecessary, but even dysfunctional, as it threatened to disrupt the social fabric and render the return to normality after the conflict more difficult. A new form of violence was introduced with the cow protection riots from the 1880s and 1890s onwards. Now the attacks were increasingly directed at sacred symbols -cows, but also mosques or temples, idols, and holy texts -threatening their desecration, if not their physical destruction. Here the violence already implied rivalling masculinities: It was the inability of the community, and mainly of the men of the community, to protect from harm what was entrusted to them (no matter whether it was the colonial state or the communal other that inflicted this harm), which was to be demonstrated or repudiated. Again, it was emotions and bodies that linked masculinities to the survival and the honourable future of the community.
These registers of violence were widespread in the late nineteenth and early twentieth centuries, and dominated the Shahabad riots in 1917. This explains why the patias mandating the attendance of the villagers were couched in such explicitly sexual terms: What was at stake was the virility of the community, revealed in its willingness and ability to protect the cows. But the Shahabad riots also showed first signs that women were to be included among the sacred objects to be protected or attacked, taking the violence to a new level: 'Mosques were defiled and Korans torn up and trampled in the mud. Women who could not escape or hide were often stripped naked, and in some cases outraged by the rioters.' 75 Though the British officers were inclined to believe these reports, the charges were difficult to link to specific instances, as 'many would rather die than make public the facts of such outrages.' At Piru, 'at the beginning of this month, the women came out in a body and stood with heads bent, weeping but said nothing.' The Magistrate 'felt what this was intended to convey. His inquiries have left no doubt in his mind that many outrages were committed.' 76 Until the Delhi riots in 1924, violence against women, or even rumours of such violence, were not publicly mentioned, though the region saw almost 40 major and minor riots during these years. In 1924, attacks against women were still so rare that they sparked off intense outrage and shame. Hakim Ajmal Khan, in his declaration to the press quoted above, exclaimed: 'Of all the incidents which form part of the recent disturbances in Delhi, to me the most humiliating and heart rending are the atrocities and cowardly assaults committed on women. . . . Some of those who claim to be the votaries of Islamic faith, not content with attacking a Hindu temple and breaking the idols, perpetrated cowardly assaults on women and children. I tremble with the deepest indignation at the very idea of my co-religionists exhibiting such wanton and callous disregard for the honour and sanctity of womanhood.' 77 Again, after the event, this form of violence almost disappeared for a couple of years, though the number of communal and other riots multiplied. In Kanpur in 1931, violence did not occur only against sacred symbols, temples, and mosques, which were attacked from the first day, and against men, but also against women and children. Autotelic violence and wanton cruelty by far surpassed the ordinary level that people had come to expect by the 1920s. However, while women and children became victims alongside the rest of the family during acts of arson and looting, this violence was not, as a rule, sexualized. 78 It is only with the beginning of the Partition riots -Calcutta, Noakhali, and Bihar, from August to November 1946 -that a new form of violence became common, one which no longer aimed only at killing, but at the utter annihilation of the adversary. It is here that the humiliation of forced conversion and rape gained their prominence. Unlike during locative violence, the disruption of the social fabric was now intended. 79 Politicians, aiming at a showdown to establish Pakistan or prevent its establishment once and for all, wanted to move beyond debates and constitutional means and worked towards a violence geared at transcending all boundaries. This created the atmosphere of an impending Armageddon, which would mark the beginning of a future different from all former futures. The numbers involved rocketed: more than 5,000 killed during the Great Calcutta Killing, 80 another 5,000 in Noakhali and between 5,000 and 8,000, with over 70,000 refugees, in Bihar. 81 The aim to create a point of no return also explained the cruelty of the violence, which now seemed to take its cue from the most outrageous rumours that had circulated previously. For these days it is hardly possible to distinguish between rumours as an exaggeration (or even an invention) of facts meant to trigger retaliatory violence and rumours as an instruction for the performance of violence.
The link between forms of violence and the body was no historical constant, but changed profoundly in the decades before Partition. What has hardly been investigated in this context is the role played by weapons as material objects and the relations they implied to the bodies. Already the three most common weapons implied quite different ways of inflicting violence. Lathis, thick bamboo sticks, were readily available and carried along on multiple occasions for self-defence against animals and robbers, but also as ceremonial weapons during processions and as tools of aggression. They were useful only in a body-to-body fight, presuming a close encounter with the adversary. But as they were so common, they did not necessarily indicate premeditation. Knives share some of these characteristics, but were more commonly associated with stray assaults at the spatial or temporal margins of a riot, as they allowed to kill quickly in a single movement (as opposed to a battering with lathis). The third weapon was brickbats, which were thrown, either by two groups at the same level or, more commonly, from rooftops, at processions or at attackers. Only in exceptional cases did rioters spontaneously find these missiles in sufficient quantities on their rooftops, usually this involved prior collection and storage and indicated a certain premeditation. Unlike the other two, this weapon did not force the adversaries into bodily contact. This form of violence was supplemented from the late 1930s by the increasing availability of firearms and the use of country-made bombs. 82 As often, these findings raise more questions than they answer. We have many sources exhorting men to perform violence and we have ample evidence of the forms this violence took. What we are lacking is the smoking gun, the ego document noted down soon after the riot, describing the emotions felt while throwing brickbats or wielding a lathi. But this might matter less for gaining an insight into the experience of felt emotions than is usually assumed. We have the sources that draw attention to different forms of violence and sensory experiences. If we assume that materiality matters, then the rich information we have on the sensorium of the riot -the sounds, sights, and smells and the encounters between bodies -can be brought into conversation with the texts premediating the event. Together, they contributed to forming and engaging specific passions, which can be identified and analysed. While we are not able to claim with certainty that a particular person experienced these emotions at a particular event, attention to these different aspects does allow us to claim with high probability that certain emotions were felt -not by everyone, and not on every occasion, but often enough by a sufficient number of people to warrant our attention and allow us to interpret riots in a new way.

Conclusion
Riots are a well-researched topic in South Asian history and, of course, the link between violence and emotions has not escaped the notice of earlier researchers. The history of emotions, this article has argued, has the potential of intervening at three different levels. First, at the level of crowd violence, it questions the facile link between bodily proximity and emotional contagion. Both space and time matter and change the shape a crowd may take. Rather than assuming that the crowd coming together as a crowd already provoked certain emotions, it has looked at the spatial layout of the cities, in order to distinguish between the scenes of the riot taking place in the maidan or the broad avenue from those enacted in the smaller lanes, as well as the microtemporalities of the riots, and has shown how emotions were at play at all stages of the riot, albeit in different modalities.
Second, the article has underlined that violence during the riots was powered by many more emotions than just anger and fear, usually associated with it, and that these emotions were historically contingent. In the period under investigation, josh had become a central category. Long before the individual riots, potential participants were already introduced to the interpretation of overwhelming and uncontrollable emotions as a sign of vitality and virility, indicating the continued survival of the community, and learned to desire them. This did not eclipse the other feelings, but has made appeals for the disciplining of emotions more ambivalent. Josh as a desired emotion is not a transhistorical phenomenon -it is hard to find before the 1870s, reaches its heyday in the period bookended by the two wars, and loses it traction in the Nehruvian era. Whether the present revival of the discourse on hurt sentiments also implies a re-appreciation of josh remains an open research question.
Third, the article has argued against a tendency to view violence itself (in distinction to its causes and effects) as a phenomenon marked by bodies and emotions and therefore outside of history. The historicization of emotions allows us to create a middle ground, which neither negates those qualities of violence that remain unsayable and resist interpretation and meaning, nor condemns researchers to a silence that might contribute to the reinforcement of the myth at the core of crowd psychology: viewing violence and riots as a primordial force. Historicized emotions go beyond interests and rational communication -but they are not beyond the scope of our inquiries.