The stampede at a Durga Puja pandal in Kolkata took the lives of 19 people on October 17. As is the case with most stampedes in India, most of the victims were women and children. The responsibility for mismanagement of crowd arrangements, which is a logical inference in any stampede, is being investigated. Some of the circumstances of this tragedy—such as the takeover of the pandal by a BJP-affiliated group and the redeployment of security personnel to Delhi at a time when they are needed in Bengal for the Navratri period—lend themselves to political disputation but are not useful to understand stampedes as a phenomenon.

Stampedes have been a recurring form of disaster in India in recent years. During a period of less than two and a half years, eight stampedes have taken place at different locations: a religious assembly in Hathras, UP; Maha Kumbh in Prayagraj; New Delhi railway station; a ticket counter in Tirupati; a theatre in Hyderabad; a political rally in Tamil Nadu; a stadium in Bengaluru; and now a puja pandal in Kolkata. The toll varied from one (Hyderabad) to 121 (Hathras), and the triggers were different, but this run of bad events can help to develop a typology of stampedes in India, which can help agencies and organisers design preventive plans.

Religious gatherings, mass celebrations and political/celebrity events figure prominently in India’s stampedes. So that’s one factor crowd management agencies need to give extra weightage for, not just while giving permission but while monitoring the event as well. From a crowd psychology point of view, such events are likely to generate sudden excitement or unpredictable movement. Crowd management agencies must revise their manuals to include special safeguards and protocols for such events.

Another red flag in these tragedies was the underestimation of crowd numbers. The organisers understated the likely crowd size when they applied for permission or were surprised by the numbers that turned up. They skimped on some things they thought were not likely to be needed. In the theatre stampede in Hyderabad, guards were arranged to protect the star but crowd management was left to outnumbered ushers. It would be wise for the licensing authorities to insist on higher provisions than currently required, including exit points, ambulances, first-aid stations, first responders, drinking water and safety monitors.

Another ingredient in the stampede mix is the trigger for panic, which varies but can be anticipated and factored for the event plans. In Hathras, it was the rush, against the direction of crowd flow, to seek a religious leader’s blessings. In the New Delhi railway station stampede, it was the anxiety of missing a train; in Kolkata, it was a sudden shower, which sent people scurrying for cover. Had the authorities worked with the organisers on the number of and distance to exit points and had both monitored crowd flow and density, the tragedies could have been averted.

Because of the recent frequency of stampedes, there is a good case for a total overhaul of crowd management technology and protocols in India. Instead of making plans by crowd numbers, we need to switch to crowd density, for which there are sensors that can be deployed as per need. Another important step is to make crowd control a collaborative effort by organisers and multiple public agencies, not just the police.