Bali: animal sacrifice in Hinduism, why it persists, and the tradition's own long argument against it

At Dakshinkali, about twenty-two kilometres south of Kathmandu, uncastrated male goats and cockerels are offered on Tuesdays and Saturdays, once in the morning and again in the afternoon. At Kalighat in Kolkata, goat sacrifice belongs to the temple's ordinary round rather than to a festival exception, and the numbers climb at Kali Puja. At Kamakhya in Guwahati, goats and pigeons are offered. None of this is folklore or a survival from a vanished century. It is Hindu practice, performed by priests, inside temples that receive millions of pilgrims a year.

So the plain answer is yes. Some Hindus sacrifice animals. It happens in Shakta worship, directed almost always to a form of the Goddess, on days the temple calendar fixes years in advance. Any defence of Hinduism that opens with "we do not do that" loses to a nine-second video search, and drags every true sentence beside it down as it falls.

The second fact deserves more attention than it gets. The case against animal sacrifice is not a modern import. It was made inside the tradition, in the Upanishads, in the Mahabharata, in the Gita, in Tamil ethical poetry, and in the practice of temples that quietly began offering a gourd where a goat had been. Hinduism does not hold one position on bali. It holds an argument, and the argument is old.

Where the practice is actually current

Bali is regional and it is Shakta. The living centres are Assam, West Bengal, Odisha, Jharkhand, parts of Bihar, pockets of Himachal Pradesh and Uttarakhand, and Nepal, where it is far more visible than anywhere in India.

The calendar matters, because outside it these temples look like any other temple. In Bengal the concentration falls on Durga Puja, particularly the sandhi puja that spans the joint of Ashtami and Navami, and again on Kali Puja. In Odisha the Chhatar Jatra of Maa Manikeswari at Bhawanipatna draws animal offerings around Maha Ashtami. In Nepal the peak is Dashain, when the killing is a public and domestic event rather than a temple speciality. Dakshinkali and Kalighat also work a weekly rhythm, with Tuesday and Saturday treated as the Goddess's days.

Set against the whole of Hindu practice, this is a minority. The great Vaishnava temples do not do it: Srirangam, Tirupati, Nathdwara and Guruvayur run on cooked food, flowers and water. Most South Indian agamic temples exclude it from the sanctum by their own ritual rules, even where a village shrine nearby performs it. Arya Samaj rejects it, Lingayats reject it, the Swaminarayan and Chaitanya lineages reject it. Not a Hindu norm, then, and not a Hindu embarrassment to be hidden, but the practice of specific communities in specific places on specific days.

The boundary is less tidy than either camp would like. Inside the Jagannath temple compound at Puri, the most famous Vaishnava establishment in India, the shrine of Bimala receives goats during Durga Puja. Odia temple literature records the ritual under its own liturgical name, Chheda Paka Niti: two male goats on each of Saptami, Ashtami and Navami, killed near the Bakula Pindi before dawn and finished before Jagannath's sanctum opens for the morning, with fish from the Markandeya tank cooked for the Goddess. Writers on the shrine, Surendra Nath Dash and Dukhi Shyam Behera among them, record that Bimala is served purely vegetarian food the rest of the year. That account rests on Odia temple scholarship rather than on any statement by the temple administration. One compound, two liturgies, and a schedule arranged so that they never meet.

Calling those communities backward is false and useless. Kamakhya is among the most learned Tantric establishments in India, and Kalighat sits in the middle of Kolkata. Any objection worth making must be made to people who have heard it already.

What the offering is understood to be

The English word "sacrifice" carries a transaction: something is given up so that something is obtained. The Sanskrit bali is closer to tribute or share, the portion set aside for a power. Inside Shakta ritual the killing is not understood as feeding a hungry deity. The animal is consecrated with mantra, the head is offered, and the body returns to the community as prasada, which is why bali and the feast are inseparable in Bengal and Assam.

The reasoning turns on substitution. The animal stands for the offerer. What is meant to die on the block is the sacrificer's attachment to a separate self, and the pashu (which means both animal and, in Tantric usage, the bound and unenlightened person) is that self made visible. This is not a modern apologetic gloss. Suchitra Samanta, writing from fieldwork in Bengali temples and homes, called the victim the "self-animal" in a 1994 study in the Journal of Asian Studies, and found the rite organised around the idea that what the Goddess consumes is the offerer.

The Kalika Purana, the Assamese text usually dated to the tenth or eleventh century that supplies the ritual charter for Kamakhya, makes the same logic explicit. Chapter 67, the Rudhiradhyaya or blood chapter, lists what may be offered: birds, tortoises, fish, goats, boars, buffaloes, deer, and, at the far end of the list, a human being and the worshipper's own body and blood.

That list needs stating rather than skipping. The Rudhiradhyaya does discuss human offering, hedged, as the scholarly summaries of the chapter describe it, with conditions that make it nearly impossible: consent, a prince's sanction, war or imminent danger, exclusion of Brahmins and of the unwilling. No temple performs it, and it is murder under Indian and Nepali law. But a reader who finds the passage independently, and finds this article avoided it, will trust nothing else here.

Texts of this kind are charters, not transcripts. The Kalika Purana prescribes offerings Kamakhya does not perform, and Kamakhya performs details the text never mentions. Hugh Urban's fieldwork in Assam and June McDaniel's ethnography of Bengali goddess worship find the same gap, and Xenia Zeiler, in the 2019 International Journal of Hindu Studies issue on sacrifice and Tantra, follows blood offerings from late medieval texts into ritual forms the manuals never anticipated.

The argument against it was made in Sanskrit first

The Mundaka Upanishad does not call sacrifice cruel. It calls it insufficient, which from the Upanishadic side is worse.

plava hy ete adrdha yajna-rupa

"Frail indeed are these boats of sacrifice, in which is held the inferior work performed by the eighteen. Fools who welcome this as the highest good go again and again to old age and death."

Mundaka Upanishad 1.2.7, rendered here from the Sanskrit rather than quoted from a single published translation. The "eighteen" are identified in Shankara's commentary as the sixteen ritual officiants together with the sacrificer and his wife.

That is an attack on the whole ritual economy, written by people who knew the rites from the inside. The Brahmana literature does prescribe animal offerings, the pashubandha and the horse sacrifice among them, and pretending otherwise is not available to anyone who has read it. The Upanishads answer that literature rather than deny it. The boat floats. It will not cross.

The Gita takes a different route to the same place. Asked in 9.26 what he accepts, Krishna sets the bar at the level of a poor household with a garden: patram pushpam phalam toyam, a leaf, a flower, a fruit, water, offered with devotion, and he takes it. Nothing there requires an animal, a fee, a priest or a fire, and it is spoken by the deity rather than about him.

The Mahabharata stages the dispute as a courtroom scene. In the Shanti Parva, at section 338 of Ganguli's translation (the critical edition numbers it differently), the gods and the rishis fall out over the word aja in the sacrificial injunction. The gods say it means the goat. The rishis say the Sruti means seeds, grain three years old that will no longer sprout. King Uparichara Vasu is asked to judge, sides with the gods, and is cursed by the Brahmanas to fall from heaven, sinking into the earth before Vishnu has him raised again. It is a scriptural argument about scripture, preserved inside scripture.

Tamil ethics reached the same conclusion without the Vedic frame. Tirukkural 259, in G. U. Pope's rendering: "Than thousand rich oblations, with libations rare, / Better the flesh of slaughtered beings not to share." The Bhagavata Purana tells it as a story. In the ninth chapter of the fifth canto, dacoits seize the silent sage Jada Bharata to offer him to Bhadrakali, and the Goddess comes out of her own image, takes up the chopper and destroys the men holding it. In her own temple, in her own text, she refuses the offering.

Assam is the sharpest case, because both positions are native to the same soil. Kamakhya continued. From the late fifteenth century Srimanta Sankaradeva built Ekasarana Dharma across the same valley, and its rejection of blood offering was a defining commitment. Maheswar Neog's history of Sankaradeva and his times sets that rejection against a religious landscape in which animal and even human sacrifice were commonplace.

The gourd, the coconut and the flour animal

The substitutions are the part of this story that most Hindus have witnessed without recognising it. In Bengali households the sandhi puja offering is an ash gourd, the chalkumro, sometimes fitted with four sticks for legs and a smaller one for a tail, then struck through with the same blade and the same mantra used for a goat. Sugarcane, cucumber, banana and mashbhakta appear on the same list, and coconuts are used across India for the same purpose. In Sanskrit ritual manuals the substitute has a technical name, pishtapashu, the animal made of flour, and by the medieval period prayoga handbooks were recommending it in place of the living victim.

Notice what substitution does not do. It does not cancel the rite, apologise for it, or import a foreign ethic. The gourd is severed at one stroke, the priest is the same, the mantra is unchanged, and red kumkum water stands in for blood. The logic that permits the swap was there at the start: if the victim always stood for the offerer, what the ritual requires is a stand-in, and the stand-in can be vegetable.

Substitution is not the tradition being reformed. It is the tradition finishing its own sentence. The rishis in the Shanti Parva argued that aja meant seed; the Bengali priest who severs an ash gourd at Ashtami has settled the argument in their favour, using their reasoning, inside their ritual.

Which is why substitution is stronger ground than denial. Denial requires pretending Kalighat is not there. Substitution can point at Kalighat and at the ash gourd three streets away and say that both are Hindu, that Hindus have argued about which is right for two thousand years, and that the argument has been moving in one direction for most of it.

Gadhimai, and what an announcement is worth

The Gadhimai festival at Bariyarpur in Nepal's Bara district, held every five years, is the largest ritual killing of animals in the world. Its founding story has the goddess appearing to a prisoner named Bhagwan Chowdhary and promising prosperity in exchange for blood, animal blood substituted for the human offering she first asked for. Humane Society International put the 2009 toll at around 500,000 animals and the 2014 toll at around 250,000. Neither figure rests on an independent count.

In July 2015 Humane Society International and Animal Welfare Network Nepal announced that the temple trust had ended the sacrifice, and the announcement travelled worldwide. Within days the temple's then chairman, Ram Chandra Shah, told the BBC that no such decision had been taken, and that devotees could be requested not to offer animals but could not be compelled. In 2016 Nepal's Supreme Court directed the government to begin phasing out the practice. When the killing went ahead anyway, campaigners filed a contempt case that remains unresolved.

The 2019 festival happened. So did the 2024 festival, at which Humane Society International counted 4,200 buffaloes beheaded inside the arena from the entry receipts, with thousands of goats and pigeons killed outside it. That distinction between the arena and everything around it explains most of the contradictions between published figures.

The totals are contested. For 2019, Humane Society International put the figure near 250,000, the BBC reported estimates of 30,000 to 40,000, and aerial footage of the buffalo enclosure suggested a little over 3,200 buffaloes against the roughly 15,000 the temple committee had expected. Advocacy groups and temple authorities have opposite reasons to prefer a high or a low number, and no neutral census exists. What is not in dispute is that the killing continued after the announcement that it had ended.

A practice embedded in a five-year vow cycle, in the household economy of the plains, and in an idea of what the Goddess is owed does not end because a trust issues a statement. Where sacrifice has receded in South Asia, it has receded the slow way: a lineage takes Vaishnava initiation, a priest accepts a gourd, a family stops sending a goat and nobody announces anything.

What the law does and does not say

The legal position is routinely stated wrongly in both directions, so here it is, slowly.

There is no national ban. The Prevention of Cruelty to Animals Act, 1960 contains section 28, which provides that nothing in the Act makes it an offence to kill an animal in a manner required by the religion of any community. Restriction comes from state legislation. Andhra Pradesh has prohibited sacrifice in places of public religious worship since its Act of 1950, carried forward into Telangana; Karnataka's Act dates from 1959, Puducherry's from 1965, Kerala's from 1968, Gujarat's from 1972 and Rajasthan's from 1975. The wording is close to identical across them: no sacrifice inside a place of public religious worship or its precincts.

Tamil Nadu moved the other way. Its 1950 Act, inherited from Madras, was a dead letter for more than half a century until August 2003, when the Jayalalithaa government abruptly instructed collectors and police to enforce it. The backlash was rapid, and the Act was repealed by Tamil Nadu Act 20 of 2004. The anthropologist Anthony Good has documented that sequence, including why a government would enforce a fifty-year-old statute and then reverse itself within months.

The court decisions have been the most reported and the least understood. On 1 September 2014 a division bench of the Himachal Pradesh High Court, Justices Rajiv Sharma and Sureshwar Thakur, prohibited animal sacrifice at temples across the state. On 10 April 2017 the Supreme Court, hearing appeals against that judgment, passed an interim order to the effect that any slaughter connected with the matters before it should be done in an area set up in accordance with law, with the municipal authority to ensure compliance. Sacrifice resumed. In September 2019 the Tripura High Court banned sacrifice at the Tripura Sundari temple at Udaipur and elsewhere in the state; the Supreme Court stayed that order in November. In 2020 the Orissa High Court restrained sacrifice at the Chhatar Jatra in Kalahandi, and in October the Supreme Court stayed that too. In February 2025 it issued notice to the Himachal Pradesh government on a petition by the Animal Welfare Board of India and the activist Gauri Maulekhi over sacrifices at a Bhunda Mahayagya held at Rohru that January.

The pattern is clear enough. High Courts have restricted, the Supreme Court has repeatedly stayed rather than affirmed, and the Himachal appeals have not been decided. Anyone who says the practice is banned in India is wrong, and anyone who says the courts have upheld it is equally wrong. The position is state by state, and where it is being litigated the most recent word is an interim order rather than a judgment.