Understanding Gujarat Violence
Published on: Mar 26, 2004

Ashutosh Varshney is associate professor of political science at the University of Michigan, Ann Arbor and director of its Center for South Asian Studies. His most recent publication is Ethnic Conflict and Civic Life: Hindus and Muslims in India (Yale University Press, 2002), which will be released in paperback by Yale in April 2003. Research for the book was partly funded (1993-5) by the SSRC through its MacArthur Program in International Security and Peace.

At 7:42 A.M. on February 27, 2002, Sabarmati Express pulled into the train station of Godhra, a small town in the Western Indian state of Gujarat, ruled by a Hindu nationalist government since 1995. What exactly happened at the train station and soon thereafter remains trapped in different narratives. Some details can, however, be reconstructed with sufficient assurance.

Sabarmati Express was carrying cadres (karsevaks) of the Hindu right from Ayodhya, where they had gone to express their vigorous support for building a Ram temple at a legally and politically disputed site.1 At Godhra, apparently, an altercation took place between Hindu activists and some Muslim boys serving tea at the train station.2 As the train began moving after its scheduled stop at the station, the emergency cord was pulled. As a result, the train stopped in a primarily Muslim neighborhood where, according to credible press reports, it was attacked by a Muslim mob. Two carriages were burned,3 and the firefighting efforts hampered. The fire killed 58 passengers, including many women and children.

A retaliatory bloodbath followed in many parts of the state. Hindu mobs torched Muslim homes and businesses, killed Muslim men, women and children, and erased mosques and graves. Instead of isolating those Muslim criminals who attacked the train and punishing them legally, as any law-bound and civilized government would do, the state government allowed revenge killings. Over a thousand lives, possibly many more, were lost over the next few weeks. Over 100,000 Muslims were pushed into the state’s ramshackle refugee camps, where basic amenities were minimal and living conditions abysmal.

Hindu-Muslim riots are not uncommon in India, but Gujarat violence plumbed new depths of horror and brutality and has come to acquire a double meaning. It was a bruising embarrassment for anyone who believes in the pluralistic core of Indian nationhood, a view enshrined in India’s constitution, a view that gives an equal place to all religions in the country, privileging none.

Hindu nationalism, India’s Hindu right, reads Gujarat violence differently. It believes in an India dominated by its majority community, the Hindus. All other religions, it has always argued, must “assimilate” to India’s Hindu core, accepting in effect that the Hindus are the architect of the Indian nation and also its superior citizens. For Hindu nationalist ideologues, the anti-Muslim violence was an ideological victory. In a formal resolution, the RSS, the ideological and organizational centerpiece of Hindu nationalism, said: ‘‘Let the minorities understand that their real safety lies in the goodwill of the majority.’’4 Laws alone, the RSS implied, as it always has, cannot protect India’s minorities.

Such views, of course, can be expressed in a democracy that protects free speech. The crux of the matter lies elsewhere. Press reports make it plausible to argue that the anti-Muslim retaliation was significantly abetted, if not demonstrably sponsored, by the elected Hindu nationalist government of the state. Were Gujarat killings pogroms, not riots? Has independent India had any other pogroms before? And what are the implications of such violence for our understanding of the role of the state in ethnic or communal riots? These are the critical issues raised by Gujarat violence.

Riots or Pogroms?

In one respect, the violence in Gujarat followed a highly predictable pattern. A recent time-series constructed on Hindu-Muslim violence had already identified Gujarat as the worst state, much worse than the states of North India often associated with awful Hindu-Muslim relations in popular perceptions.5 It had also specified three Gujarat towns—Ahmedabad, Vadodara and Godhra—as the most violence-prone: these three turned out to be the worst sites of violence in March and April 2002. It was also argued that the outbreaks of communal violence tend to be highly locally concentrated: many towns, only a few miles away from the worst cities, have insulated themselves from communal riots, entirely or substantially. In contrast to Ahmedabad, Surat’s old city (not the part where its shantytowns are) was argued to be such an example: yet again in March and April 2002, the violence in Surat was minimal, even as Vadodara and Ahmedabad, neither too far away from Surat, experienced carnage. Not everything about Gujarat violence was, however, entirely predictable. In one respect, the violence was shockingly different. Unless later research disconfirms the proposition, the existing press reports give us every reason to conclude that the riots in Gujarat were the first full-blooded pogrom in independent India.

According to dictionaries, a pogrom means:

“An organized, often officially encouraged massacre or persecution of a minority group, especially one conducted against Jews.” (www.dictionary.com)

“A mob attack, either approved or condoned by authorities, against the persons and property of a religious, racial, or national minority.” (www.britannica.com)

Reports in almost all major newspapers of India, with the exception of the vernacular press in Gujarat, show that at least in March, if not April, the state not only made no attempt to stop the killings, but also condoned them.6 That the government “officially encouraged” anti-Muslim violence—something often believed—cannot be conclusively proved on the basis of the evidence provided by newspaper reports, though later research may well prove that. What is unquestionable is that the state condoned revenge killings.

The statements of non-governmental organizations most closely associated with the state government are highly indicative. According to the chief of one such organization, the Vishwa Hindu Parishad (VHP), a leading Hindu nationalist body, Gujarat was “the first positive response of the Hindus to Muslim fundamentalism in 1,000 years.”7 The reference here is to the original historical arrival of Muslims from Central Asia and the Middle East to the Indian subcontinent, a time when a long Hindu decline, say the Hindu nationalists, also set in. On this reading, the rise of Muslims in Indian history and the Hindu decline are integrally connected, the former causing the latter, and a revenge for historical wrongs is necessary.

The Hindu right believes that its elected government did exactly what was required: namely, allowing violent Hindu retaliation against the Muslims, including those who had nothing to do with the mob that originally torched the train. For others, of course, it is not the job of the government, whatever its ideological persuasion, to stoke public anger, or to allow it to express itself violently, regardless of the provocation. No elected government that has taken an oath to protect the lives of its citizens can behave the way criminal gangs do, thirsting for a tit-for-tat. This is why Gujarat killings have been a source of bitter debate and intense agony in India.

It is sometimes suggested that the anti-Sikh violence in Delhi, after the assassination of Indira Gandhi on October 30, 1984, was the first pogrom of independent India. This argument is not plausible, for the differences are critical. To illustrate the major differences, one can do no better than cite from a most brilliant column written by a senior Indian journalist, who personally covered the 1984 anti-Sikh riots:

First of all, the ordinary mass of the Hindus in Delhi never got involved in the riots—many of us put on crash helmets, picked up hockey sticks and cricket bats, wickets, anything at night to run vigils in our streets so no “outsiders” could harm our Sikh neighbours. How many such stories have we heard from Gujarat? Second, once the government got its act together (within 72 hours) all rioting stopped, as if someone had blown the whistle and called off a game or a movie show. Third, and this is the most important distinction, there was shame, embarrassment, contrition, even fear on the faces of the top civil servants, police officers, Congressmen. They knew something terrible had happened. Rajiv Gandhi may have made his insensitive “when a tree falls earth shakes...” statement to rationalise the killings, but damage control started immediately.

....[A]s the riots were dying out on November 3 (Mrs. Gandhi had been assassinated on October 30) Delhi’s Lieutenant-Governor, P.G. Gavai, was fired. … The Station Head Officer (SHO) of Trilokpuri (police station) was removed on November 2. The police commissioner, Subhash Tandon, was replaced on November 12. So were Deputy Commissioner of Police (east), under whose jurisdiction Trilokpuri fell, Additional Police Commissioner (range), and Deputy Commissioner of Police (south). Within a month or so they were all facing departmental inquiries. Contrast this with what happened in Gujarat. Did any policeman get removed or punished for non-performance or complicity? Narendra Modi, on the other hand, moved out mainly those who had been effective, true and loyal to the uniform....

The Congressmen whose names surfaced or were even popularly mentioned in connection with the killings all paid the price. Political careers of H.K.L Bhagat, Jagdish Tytler and Sajjan Kumar never recovered from the taint of 1984 although nobody was ever convicted.... Isn’t it a bit different now when leading lights of the BJP go around talking of “Hindu consolidation,” of Modi having become a “Hindutva hero” or the likely electoral dividend of the killings?”8

The larger point should be clear. Because of their intense anti-Muslim ideology and a Hindu conception of the nation, the leading Hindu nationalist organizations, such as the VHP and RSS, have celebrated the anti-Muslim violence as an act of nationalism. In contrast, the Congress party never developed an anti-Sikh ideology. This should explain why the Congress ended up developing an attitude of contrition, but the VHP, deeply intertwined with the state government in Gujarat, found hacking and burning Muslims after the Godhra provocation a celebratory and ideologically correct act. It is the latter which makes Gujarat riots a clear pogrom. There is no contrition yet in the statements of the Gujarat state government, or of leading Hindu nationalist organizations. The anti-Sikh violence of 1984 was significantly different.

In the Gujarat government’s eyes, Muslims are disloyal and deserve to be treated harshly, regardless of whether all Muslims were involved in, or supported, the torching of train at Godhra. No distinction need be made between Muslim criminals and innocent Muslim citizens. And the most powerful civil society organizations—the VHP and RSS—are also of the same view. Instead of civil society resisting the state, or the state resisting marauding civic groups like the VHP, there was a coincidence between the two in March 2002. It is this coincidence that created the ideal conditions for a pogrom.

Causes of Riots: Towards a Deeper Understanding

It is often said that if the state were communally neutral, or scrupulous enough to protect the lives of all its citizens, there would be no large-scale communal violence. This argument is true, but only trivially so. Moreover, it does not advance our understanding of peace.

Whenever major communal violence has taken place in independent India, academics, activists, legal experts and journalists have been intensely critical of the state. All scholars as well as judicial inquiry commissions instituted to investigate riots in independent India so far have focused on riots and violent towns, not on towns that did not explode, even while other cities were burning. Bulandshahar, next to Aligarh, and Saharanpur, next to Meerut, have rarely been infected by the communal orgy of their neighboring towns. If the researchers and judges only investigate violence, the failure of state organs in preventing riots is bound to be a foregone conclusion. There is no mystery to be unraveled here.

To understand riots better, we need, first of all, to compare systematically the episodes of mass violence with episodes of peace. It is about the only way scholars know of figuring out whether they are right about their understanding. In the absence of such comparisons, we can’t convincingly answer a fundamental question: how do we know we are right?

The logic underlying this proposition is simple, often misunderstood, and worth restating.9 Suppose on the basis of commonalities, we find that inter-ethnic economic rivalry (a), polarized party politics (b), and segregated neighborhoods (c) explain ethnic violence (X). Can we, however, be sure that our judgments are right? What if (a), (b) and (c) also exist in peaceful cases (Y)? In that case, either violence is caused by the intensity of (a), (b) and (c) in (X); or, there is an underlying factor or the context that makes (a), (b) and (c) result in violence in one case but not in the other; or, most intriguingly, there is yet another factor (d), which differentiates peace from violence. It will, however, be a factor that we did not discover precisely because peaceful cases were not studied with the violent ones.

Following this method, it can be demonstrated that dreadful though it is, Hindu-Muslim violence is neither endemic nor widespread in India.10 Rioting is heavily concentrated in a handful of cities and towns. In thousands of villages and towns, Hindus and Muslims manage to live together. There may be tensions and small clashes, but they do not degenerate into riots. Since horrific acts of rioting make news and quiet conduct of everyday life does not, we get a distorted picture of the extent of violence and killings.

The share of villages in communal rioting has been miniscule. Between 1950-95, rural India, where more than two-thirds of Indians live, accounted for a mere 4 percent of all deaths in Hindu-Muslim riots. Such rioting is primarily an urban phenomenon. Moreover, within urban India itself, eight cities accounted for about 46 per cent of all deaths. These riot-prone cities have less than a fifth of India’s urban population and a mere 6 per cent of the country’s total population, urban as well as rural.

What explains these local concentrations of communal violence? Based on a comparison of six Indian cities, three riot-prone and three entirely or mostly peaceful, my book (see endnote 5) argues that the pre-existing local institutions of civic engagement—political parties; business associations; trade unions; professional associations of teachers, students and lawyers; NGOs; reading and film clubs, especially if they are mass-based, as in South India—explain why some towns remain peaceful, while others go up in flames.

Periodically, provocative events do take place. Such events can be likened to “sparks.” Where Hindus and Muslims are integrated in local civic organizations, sparks get extinguished. In towns where Hindus and Muslims are segregated and no common civic sites exist, sparks can easily turn into conflagrations, consuming tens and hundreds of lives. This was as true in March and April of 2002 as it has been for many decades, including during partition, unquestionably the worst period of Hindu-Muslim violence.

Such local concentrations of violence suggest that despite the Hindu nationalist claims about Muslim attacks on Hindu culture in the past, only some Hindus, and only some places, allow such perceived “historical and national wrongs” to undermine the political, economic and social links between communities. Privately, they may well have anti-Muslim feelings, but such feelings are not allowed to turn into violence.

Since independence, only two regions of India—the North and the West—have had repeated rioting and, within these regions, only some cities and towns. Despite having a substantial Muslim population, southern India, with the major exception of the city of Hyderabad, has had remarkably low levels of Hindu-Muslim violence. The same is true of eastern India over the last few decades. Hindu-Muslim divisions, the cornerstone of Hindu nationalism, have not been a central feature of the politics of the South and the East. Scholars have tended to draw too many unacceptable generalizations from the North or the West about the whole country. In most parts of India, the relationships of Hindus and Muslims are so extensive—often in organizations and associations—that the Hindu nationalist dream of a nationwide Hindu-Muslim cleavage, from which they would politically benefit, is most unlikely to become a reality.

What is a Better Bet for Peace: The State or Civil Society?

Given the arguments above, how should we conceptualize the role of the state in moments of communal violence?11 After more than 50 years of experience with state behavior during Hindu-Muslim riots, we should revise our view of what the state does in such moments, why it does so, and how to make the state behave better.

Here are four proposals:

First, on the major fault-lines of a polity, as Hindu-Muslim relations are in India, the state tends to act in a politically strategic, not a legally correct, manner. This is true in much of the world. Consider the Sri Lankan state on Sinhalese-Tamil relations, the Malaysian state on Malay-Chinese relations, and the US on race relations, though the US is beginning to come out of its racial bind. States should not act in this manner, but they do, combining legality, morality and political calculations in an unpredictable way. The state in Gujarat acted in a primarily strategic, not a legally correct, manner.

Secondly, this more realistic understanding of how states function neither means that citizens should cease to criticize and pressure the state when it fails to protect lives in riots, nor that they should stop trying every constitutional means of punishing the state. But, while making every attempt to put pressure on the state, they should not bet on it to rectify its behavior any time soon. If, as a result of such criticism and pressure, the state corrects itself on major fault-lines, it should be viewed as a happy low-odds outcome of citizen activism.

Thirdly, working on, and building, integrated civic networks is a better bet. Towns where Hindus and Muslims continue to be integrated—in businesses, political parties, unions, professional associations of lawyers, teachers, doctors and students, and clubs—are also towns where riots remain either absent or rare. An integrated organizational and civic life makes the state behave much better than intellectual and political exhortations that it do so. This remains one of the more important findings of my study.

Finally, it follows that citizen action has to take two forms: (a) while continuing to pressure the state for its dereliction of constitutional duty, it should (b) focus on building integrated civic structures. The first has been the primary strategy for citizen action, in India and elsewhere, thus far. Such action is necessary, but not sufficient. Citizen initiatives should follow a two-track strategy, combining (a) and (b). The state, otherwise, will continue to get away with its utter misconduct and gross disrespect for human lives during ethnic riots.


1 A Muslim mosque once stood at this site; Hindu militants razed it in December 1992.

2 It is believed in some quarters that the abducting of a young woman by Hindu militants traveling in the train triggered the attack on the train. Careful analysis shows that this story is simply wrong. For the most conclusive refutation of the story, see Prem Shankar Jha, “The Mystery Email,” Outlook (Delhi), March 25, 2002.

3 While initial reports indicated the flame came from the outside, forensic evidence indicates strongly that the fire began within. The Gujarat Forensic Science Lab (FSL) found that at least 60 liters of petrol were inside the carriage before the fire was set. Some eyewitnesses reported seeing individuals carrying jerry cans filled with petrol enter the train.

4 “Majority’s Goodwill Vital: RSS,” The Times of India, March 26, 2002. Born in the 1920s, RSS stands for the Rashtriya Swayamsevak Sangh (National Volunteer Corp). It is the parent organization of all Hindu nationalist organizations, including the political party, BJP, which at the time of this writing has ruled the state of Gujarat since 1995 and has ruled Delhi in a multi-party coalition since 1998.

5 Constructed jointly by Steven Wilkinson (Duke University) and me for the period 1950-95, the basic statistical results of the dataset, summarized below, are reported in Ch. 4 of my Ethnic Conflict and Civic Life: Hindus and Muslims in India (Yale University Press, 2002).

6 This is based on a close reading of the following newspapers: The Times of India; The Indian Express; The Hindustan Times.

7 “Parivar Wars,” The Times of India, June 26, 2002. VHP stands for the Vishva Hindu Parishad (World Hindu Council).

8 Shekhar Gupta, “Pot is Blacker than the Kettle,” The Indian Express, April 6, 2002.

9 For an elaborate argument for the need for variance in social science research, see Gary King, Robert Keohane and Sydney Verba, Designing Social Inquiry (Princeton University Press, 1993). For why discovering commonalities may matter even in a world where variance exists, see Ronald Rogowski, Symposium on “Designing Social Inquiry,” American Political Science Review, 89 (June 1995).

10 These observations and the figures below are based on Chapter 4 of my book.

11 I first presented these arguments in an article, “Making the State Behave, and Well,” The Indian Express, April 23, 2002.