The resurgence of the Ayodhya agitation and the violence surrounding it raise two important questions. What was the pattern of violence during the previous turbulent phase of the Ayodhya movement? And what might one expect now?

The 1989-92 period can legitimately be described as the peak of the Ayodhya agitation. Whichever way one looks at it-national, state or local-it was the worst period of communal rioting in India since Partition. The death toll in 1992 was over 1,200, the highest during 1950-95, and arguably till today. The previous high was in 1969, primarily as a consequence of the horrific rioting in Ahmedabad.

The worst affected states during 1989-92 were Maharashtra and Gujarat. Mumbai led the overall casualty figures, both nationwide and in Maharashtra. In Gujarat, Surat and Ahmedabad were the worst hit. Uttar Pradesh came in third, but the fourth worst affected state, Bihar, experienced remarkably low levels of violence, partly because its politics had been decisively reshaped by Laloo Yadav's lower caste-Muslim alliance.

BURNING ISSUE: Ahmedabad bore the brunt of recent riots in Gujarat

Indeed, the otherwise infamous Bihar was the first northern state to illustrate a longstanding southern "political law": the higher the intensity of lower-caste politics, the lesser the likelihood of Hindu-Muslim violence. This inverse relationship between caste and communal politics has been true of Tamil Nadu and Kerala for many decades. Kerala is especially significant as Muslims have constituted over 20 per cent of its population for a long time. Due to the clout of their principal party, the Muslim League, Muslims have received many concessions. Yet there has been virtually no Hindu-Muslim rioting in the state. The Malayalis do not generally associate the Muslim League with communally provocative politics, only with community-based politics.

This inverse relationship suggests another important conclusion. As the significance of lower-caste politics continues to rise in Uttar Pradesh, the state's historical proneness to communal riots should decline. It will, of course, not be entirely eliminated for some time, but will progressively go down.

In the end, however, it is the city or town-level analysis, not the national or state-level analysis, that generates the most significant results for understanding communal riots. First of all, between 1950 and 1995, only about 4 per cent of all deaths in Hindu-Muslim riots took place in India's villages, where 65 per cent of the country still lives. Second, in urban India, eight cities-Mumbai, Ahmedabad, Vadodara, Hyderabad, Aligarh, Meerut, Kolkata and Delhi-accounted for nearly half of all deaths. A few towns in each state drove the state-level aggregate figures. Only two cities, Ahmedabad and Vadodara, accounted for nearly 80 per cent of the total deaths in Gujarat; 88 per cent of all deaths in Maharashtra took place in the six worst affected towns of the state; and most dramatically, 80 per cent of all deaths in Andhra Pradesh took place in Hyderabad.

Hindu-Muslim violence is primarily urban and locally concentrated. The problem of communalism is much wider than the incidence of communal riots. Thankfully, violence never has a statewide or a nationwide spread. State or national politics simply provide the spark that may touch off or exacerbate rioting, but it is the local-level relationships that the Hindus and Muslims have developed that either deflect that spark or allow it to start a conflagration.

In the present phase of the Ayodhya movement, the eight worst cities listed above will have to be watched. The category of "the next worst" is also potentially inflammable and includes Bhopal, Kanpur, Jalgaon, Indore, Varanasi, Allahabad, Nagpur and Aurangabad. In Gujarat, after Ahmedabad and Vadodara, Godhra has traditionally been the third worst. Surat runs the risk of becoming a new casualty. Though historically peaceful, it had its first Hindu-Muslim riots after nearly 70 years in 1992-93. The tearing down of the Babri mosque constituted the spark last time and the sprawling shantytowns were the sites of nearly 200 deaths. Surat slums would yet again need careful watching.

One hopes the Ayodhya agitation does not scale the ferocious heights of 10 years ago. Another low turn in Hindu-Muslim relations can only weaken the country, domestically as well as internationally. Post 9/11, the world is watching how India and its majority community treats the minorities.

The writer is associate professor and director, Center for South Asian Studies, University of Michigan. He is the author of Ethnic Conflict and Civic Life: Hindus and Muslims in India, to be published in India in July by Oxford University Press.