Two counts of the same dead sanitation workers, 1.8x apart

On June 26, 2026, three sanitation workers went into a septic tank in Mundka, on Delhi’s western edge. They had no oxygen support, no gas detectors, no harnesses, no rescue plan. One collapsed. A second went in after him and collapsed too. A third followed and did not come out either. The Dalit Adivasi Shakti Adhikar Manch (DASAM), a rights coalition that has been running its own monthly count of these deaths through 2026, logged all three as part of a June toll of 18 — the worst month in a five-month tally that by June stood at 55.

Set that number beside a second one. In July 2026, Minister of State for Social Justice and Empowerment Ramdas Athawale told the Lok Sabha, in a written reply to an unstarred question from DMK MP D. Ravi Kumar, that 332 people had died in sewer and septic-tank cleaning between January 2021 and June 30, 2026 — a four-and-a-half-year span, not DASAM’s five-month one — citing data from the National Commission for Safai Karamcharis and breaking it down by state: Maharashtra highest at 58, then Haryana at 47, Tamil Nadu 38, Uttar Pradesh 37, Delhi 29, Gujarat 27, Rajasthan 23, Madhya Pradesh and West Bengal 13 each, Karnataka and Punjab 12 each, Telangana 9, Andhra Pradesh 5, Dadra and Nagar Haveli 3, Bihar and Odisha 2 each, Goa and Jharkhand 1 each. The Safai Karmachari Andolan (SKA), the national movement that has organized sanitation workers for three decades, counts 593 deaths over that same four-and-a-half-year window — a gap of 261 more dead people, by one count, than the other count acknowledges, as the chart below shows.

A gap that size is not usually a story about arithmetic. It is a story about definition. Manual scavenging deaths get contested at the boundary of the category itself: who counts as a sanitation worker doing this work, what counts as dying “in the course of” it rather than adjacent to it, and whether an informal or unregistered laborer, working outside any municipal contract, shows up in a government ledger built around formal employment records. That boundary is where a specific, well-documented arrangement does its work: municipalities and industries that need sewers cleaned overwhelmingly hire through labor contractors rather than employ workers directly, and a contractor who sends a man into a tank with no gear, no insurance, and no written contract leaves no paper trail establishing the death happened in the course of employment at all. A government commission compiling figures from official, registered incidents will structurally undercount a population whose employment was designed, from the start, to be undocumented. DASAM said as much of the Mundka deaths directly: the three men who went in one after another trying to save each other were killed by “caste violence, contractor exploitation, institutional negligence, weak law enforcement, and systemic state failure” compounding on each other. The Safai Karmachari Andolan’s higher count almost certainly reflects fieldwork and testimony reaching exactly the workers that official channel does not. Which count is “right” depends on the question being asked. If the question is “how many deaths has the state formally recorded,” 332 is the honest answer. If the question is “how many sanitation workers died,” 332 is very likely a floor, not a ceiling.

Fairness requires taking the government’s figure seriously on its own terms, not dismissing it as a cover-up by default. An official count sourced to a statutory commission is not nothing — the same reply disclosed that 193 FIRs have been registered against agencies and officials over the past five years for hazardous manual cleaning, which means at least some of these deaths are generating a legal record, even if the reply says nothing about how many of those FIRs produced a prosecution or a conviction. Civil-society tallies carry their own methodological pressures too, including the incentive to count generously when the count itself is an advocacy tool, and SKA has not published a detailed public methodology for how it arrives at 593 or which incidents make up that figure. Without that methodology in hand, it is not possible to certify SKA’s number as more accurate than the government’s, only to say the two are measuring different populations by different rules, and that the direction of the gap (undercount, not overcount, on the state’s side) is the pattern this kind of data reliably shows worldwide when the affected population is informal and undocumented. Whether “sewer and septic-tank cleaning” means an identical scope of work in both counts is also unclear, which limits how far this comparison can be pushed.

SKA’s national convener, Bezwada Wilson, did not treat the 261-person gap as a technical dispute. “Why is the government hiding 261 sewer and septic tank deaths?” he asked, calling on the prime minister to acknowledge the failure and act. SKA’s own statement went further, calling the discrepancy “a matter of national shame and an act of untouchability that instead of stopping deaths, the government is only focussed on obscuring the truth” — naming the undercount itself as a caste practice. That framing answers a question the numbers alone can’t: sanitation workers have made the same point about their own working conditions before. After an earlier Athawale statement to Parliament describing sewer cleaning as an occupational hazard rather than a caste-assigned one, Surya Ram, a Delhi sanitation worker who has cleaned septic tanks and drains for over a decade, told Outlook India in 2025: “The reality is that what we do is not just an occupation, it’s a caste-based legacy that has kept people like us trapped in unsafe, exploitative and degrading labour for generations.” The dispute over 332 versus 593 is downstream of that same disagreement about what kind of thing is even being counted.

None of this is unenforced for lack of a law. Manual entry into a sewer or septic tank has been a criminal offense since the Prohibition of Employment as Manual Scavengers and their Rehabilitation Act of 2013, and the Supreme Court told the state to enforce that ban and mechanize the work in Safai Karamchari Andolan v. Union of India in 2014. SKA’s own Delhi demonstration in late July 2026 called for the existing law to be applied — a narrower, more falsifiable demand than it sounds: a contractor or municipal official prosecuted, not an FIR filed and left open. What Athawale’s reply also disclosed, per the same report, is that the government has no plan for mechanized sewer cleaning to replace manual entry. That detail outlasts either death count. A number, however carefully defined, describes the past. The absence of a mechanization plan describes what happens next: more workers sent down by hand, into tanks like the one in Mundka, and, on the trajectory both counts describe, more deaths that the next parliamentary reply will again have to reconcile against whatever DASAM and SKA have counted by then.