Exhibit · the people of the data
A community used to model the Indus woman’s deep ancestry catches snakes for a living, and was told none of this.
In 2019 the first genome from the Indus Valley Civilisation was published — a woman buried at Rakhigarhi around 2500 BCE. Her ancestry has two parts, and the indigenous half is the one this page is about. In the formal model that half was fitted using the Andamanese, because modelling requires an unadmixed source. But among living mainland populations, two of the communities repeatedly reached for when that component is modelled are the Paniya and the Irula, a Scheduled Tribe of the Nilgiri hills — landless, historically criminalised by law, and among the poorest communities in India. This page is about what that means, and about who was never told.
Three arrivals, and who was already there.
the layers, drawn
The Irula are not the Indus population and do not descend from it in any special way. A proxy is a statistical instrument: the living group whose ancestry most closely resembles a component being modelled. It is chosen for a statistical property, and choosing it says nothing about descent.
And a Y-haplogroup frequency is not an ancestry percentage. F*-M89 at 53% describes one unbroken father-to-son line, one ancestor out of more than a thousand in the same generation. AASI proportions come from genome-wide analysis and are a different measurement. The two are routinely conflated; nothing on this page depends on treating them as the same.
What the model actually does.
Rakhigarhi, decomposed
What the paper found, and what was announced.
6 September 2019, New Delhi
| The published paper | The press conference |
|---|---|
| Iranian-related ancestry plus AASI. Zero steppe ancestry — consistent with steppe arrival after 2500 BCE. | The study "completely sets aside the Aryan Migration/Invasion Theory." |
| A genome. No linguistic content of any kind. | The Indus people spoke Sanskrit. |
| The indigenous half modelled on a Nilgiri Scheduled Tribe. | "Our premise that the Harappans were the Vedic people has received strong corroborative scientific evidence." |
Vagheesh Narasimhan and Niraj Rai — authors of the same paper — publicly stated the evidence did not support what was said from the podium. Three weeks earlier Rai had said on the record: "I know people won't be happy to hear this. But I don't think we can refute it anymore. A migration into ancient India did happen."
The findings were subsequently written into NCERT textbooks. What was suppressed was not the data. It was the sentence that would have made the data legible to an ordinary reader.
Who the proxy actually is.
the Irula, in the present tense
Irula snake-catchers supply the venom from which most of India's antivenom is produced — in a country with among the highest snakebite mortality on earth. Their tracking skill is so specific that in 2017 two Irula men were flown to Florida to hunt invasive Burmese pythons in the Everglades, and caught fourteen in under two weeks. Vadivel Gopal and Masi Sadaiyan received the Padma Shri in 2023.
A community whose expertise saves tens of thousands of Indian lives a year holds no land, and its members are still picked up by police.
The part that makes this an exhibit.
extraction, and who is told
The Irula did not learn any of this from a school textbook, a museum, or a scientist. A young Irula woman asked about it in 2026 had never heard that her community is used to model the ancestry of the Indus Valley. The finding is in Cell. It is in the ancestry-modelling literature. It has been in the news in three languages. It has not reached the people it is about.
A Harvard laboratory holds more than half the ancient human genomes ever produced. A ministry holds the bones, the country's only clean room, and the microphone — and announced the conclusion of its next study before running the samples. The Irula hold neither the data, nor the interpretation, nor a place at either table.
The shape is familiar: extraction from the body, aggregation somewhere else, analysis in a language the source cannot read, and conclusions returned as headlines that reclassify the source. What a platform does with behaviour, a laboratory does with ancestry — and here the state has positioned itself as the platform.
Indigenous data sovereignty is an established field. The CARE Principles for Indigenous Data Governance. The Havasupai settlement in the United States, where a tribe's blood samples were used for research it never consented to. Māori data sovereignty frameworks in New Zealand. The Indigenous Peoples Council on Biocolonialism called for a boycott of the Genographic Project over consent and commercial use of indigenous DNA — and the largest jāti-level genetic study of Tamil Nadu is a Genographic output.
No Adivasi data-sovereignty framework exists. No community consent protocol for Indian ancient DNA has been located. That absence is the finding. And it is not an oversight — it is the one thing the two custodians agree on.
What decolonising this would require.
and the trap in the word
"Give the people control of the narrative about their ancestry" is exactly the argument a nationalist ministry makes when it says foreign scientists have no standing over Indian bones. That is not decolonisation. It is a second extraction by a closer power, wearing the language of the first.
CARE and Havasupai are about the community that was sampled. Here that is the Irula and the Paniya — not the ministry, not the university, not the excavator. Any framework that cannot tell those apart will hand custody to the party with the most power and the least accountability to the people in the sample.
- Return the finding first. Before the press conference, before the journal embargo lifts — in the community's own language, in the settlement, to the people whose relatives were sampled.
- Consent that includes interpretation. Not just permission to take a swab, but standing to object to what is published about you.
- Benefit that is material. Antivenom built on Irula knowledge is a national medical asset. Ancestry modelled on Irula genomes is a national scientific asset. Neither has produced land, security or standing for the Irula.
- And the obvious one: if a state's textbooks are being rewritten around who is indigenous, the communities carrying the oldest lineages on the subcontinent might be in them.
The honest limits.
- Cited: the Rakhigarhi genome composition and absence of steppe ancestry (Shinde et al. 2019, Cell); the role of southern hill communities as modern proxies in ancestry modelling (Yelmen et al. 2019, Mol. Biol. Evol. 36:1628 — abstract, indexed full-text excerpts and citation statements read here; the paper’s named modern proxy is the Paniya, and no statement ranking the Irula as closest was found); the F*-M89 frequency of 53.3% in hill foragers and the significant under-representation of R1a1 and L1 there (ArunKumar et al. 2012, PLOS One, doi); the press-conference statements and the co-authors' disagreement, from contemporaneous reporting; the Irula legal and economic history from the published socio-economic literature and the cooperative's own record.
- NOT claimed: that the Irula are the Indus population, or descend from it. NOT claimed, after audit: that they are the closest surviving model for that ancestry component — that ranking was attributed to a paper which, in the text reachable here, names the Paniya as its modern proxy and treats a reconstructed haplotype set, not a living group, as the AASI proxy. What remains is that Irula and Paniya samples are used in this literature, which is a statistical relationship and not a genealogy.
- NOT claimed: that AASI proportions are settled figures. Estimates vary by study and method. After audit, the direction is not asserted either: no accessible source ranks living mainland populations by AASI proportion, and the page no longer implies one.
- NOT claimed: that the Irula finding was suppressed. It was published. What was inverted was the public account of what the paper meant.
- Not used: per-community ancestry percentages circulated by commercial DNA-testing companies. Several publish tidy figures with no sampling method or peer review.
- On the individual account: the young Irula woman's unfamiliarity with this research is a single unverifiable report. After audit it is retained only as an illustration explicitly marked as such; it is not evidence of community-wide awareness, it has no source a reader can check, and nothing on this page rests on it.
What this page could not verify
the premise it was built on, and why it is not made here
The page's founding premise was not supported by the source it was attributed to. The page now declines to make that claim, and the reason a visitor should trust the refusal is the record of what was and was not read.
Every claim, and the depth of its source
The page is about a community used as an instrument. It owes a visitor an exact account of its own evidence.