Evidence page · the other half of the record
Every caste's maternal lineages are old, local and shared. Only the men were sorted.
Almost everything written about caste and genetics is about the Y chromosome — which lineage peaks in which stratum, which arrived when. The maternal record is discussed less because it is less obliging. Across the subcontinent, from Brahmin to Scheduled Tribe, mitochondrial lineages are overwhelmingly indigenous, deeply rooted, and substantially shared. The sorting that produced modern caste is visible in one line of descent and largely absent from the other. This page is that asymmetry, and what follows from it.
The asymmetry.
two records, one population
What that looks like in South Asia.
the same population, read two ways
| Paternal (Y chromosome) | Maternal (mitochondrial DNA) | |
|---|---|---|
| External input | Substantial. Steppe-derived R1a-Z93 enriched in Brahmin and Bhumihar groups; L1 and J2 from the Iranian plateau. | Minimal. Overwhelmingly Indian-specific branches of haplogroups M and N. |
| Sorted by rank? | In places, yes — R1a above 50% in some Gangetic Brahmin samples, near-absent in hill tribes. | Largely not. The same deep lineages run under every rank. |
| Tribal vs caste | Strong differentiation between tribal groups; weak among castes. | Tribal mtDNA coalesces at Indian-specific M and N branches that span populations of every social rank across the subcontinent. |
| Gene flow into castes | Restricted by endogamy. | Higher into caste groups than into tribal ones — read in Andhra as historical upward female mobility. |
High paternal affinity to eastern Europeans through R1a1 — and maternal polymorphisms revealing primarily Indian-specific lineages. The same individuals. Two records. One says arrival; the other says the women were already here.
Three mechanisms.
why the two records diverge
Three mechanisms
| Mechanism | Effect |
|---|---|
| Male reproductive variance | A small number of men can father a very large number of children; women cannot. A militarily or politically dominant male group leaves a Y signature far larger than its actual numbers. The size of a genetic signature and the size of a migration are not the same quantity. |
| Patrilocality and female exogamy | Where women move to the husband's community and men stay, mtDNA is redistributed between neighbouring groups while Y lineages stay put. Maternal variation is homogenised regionally; paternal variation is sharpened. |
| Directional hypergamy | Where marriage rules permit women to move up in rank but not men, ancestry travels upward through women and does not travel down through men. This is the arrangement the law books devote their energy to policing. |
Three competing explanations for one pattern, best read as a comparison.
The oldest reading, and what it suggests.
two findings from the Bronze Age
The skeletal population of Cemetery R-37 at Harappa shows strong genetic affinities among the women buried there — read as indicating that women stayed and men married in. The opposite of the pattern that later becomes standard. That is a finding about one cemetery, not a society, and it is flagged here as a thread rather than a conclusion.
And the formation of the Indus population itself was not sex-neutral: the same asymmetry between the maternal and paternal lines appears in the published modelling of the Indus period, not only in what came after. Sex-biased mixing is not only a feature of the steppe arrival. It is there in the making of the Indus.
Where the other system survived.
matriliny at the margins
If patrilineal inheritance is what squeezed the male lines, then the places where it arrived last should look different — and they do. Every surviving matrilineal system in South Asia sits at maximum distance from the Gangetic plain where gotra and the law books were composed.
| Where | System | Status |
|---|---|---|
| Kerala, coastal Karnataka Nair, Bunt, Tiyya | Marumakkathayam — property through the sister's son | Dismantled by statute. Malabar Marriage Act 1896; Madras Marumakkathayam Act 1933. Colonial and then Indian law imposing patrilineal succession. |
| Meghalaya Khasi, Garo, Jaintia | Matrilineal descent, matrilocal residence; youngest daughter inherits | Intact. |
| Lakshadweep | Matrilineal | Persisted under Islam. |
Nair society had powerful men. What differed was the operating system for inheritance — descent, residence and property running through women — and that system does not produce a male-lineage bottleneck. These are not proof that pre-agricultural South Asia was matrilineal. They show what inheritance can look like without patrilineal property, in the regions the intensification reached last.
And the third layer lands here too. The inheritance system that survived three thousand years of the Gangetic order did not survive fifty years of the Madras Presidency. Colonial law finished by statute what the ancient expansion had left undone.
What this refuses.
stated plainly
- Caste groups do not descend from separate ancestral peoples. If they did, the maternal record would show it, and it does not. The deep female lineages run under every rank.
- The ranking was imposed on a population already substantially shared — and the sharing is most visible precisely in the line of descent that caste ideology cared least about controlling and most about policing.
- Whatever happened at the arrivals was done to men's reproduction and through women's. Local male lines lost; local female lines did not. That is the demographic residue of one male population out-reproducing another, by whatever combination of violence, coercion and the ordinary advantage that accompanies power.
- And the law books are late. The obsession with varṇasaṃkara — mixture — is not describing a hypothetical. It is a category built to manage the consequences of exactly this, written after the fact.
The honest limits.
- Cited: the global male bottleneck and rising female effective size (Karmin et al. 2015, Genome Research); the patrilineal-clan explanation (Zeng et al. 2018; Guyon et al. 2024, Nature Communications); Indian-specific M and N branches spanning ranks, and the eastern Indian Brahmin paternal/maternal contrast (Kivisild et al.; Basu et al.); the Andhra mtDNA gene-flow pattern (BMC Genetics 9:86); the R-37 female affinity finding; steppe ancestry as male-derived and enriched in Brahmin and Bhumihar groups (Narasimhan et al. 2019); the Nair inheritance statutes.
- NOT claimed: that pre-agricultural South Asia was matrilineal. The surviving systems show possibility, not history.
- NOT claimed: that women were unaffected by caste. Their lineages were not eliminated; their lives were governed in detail, and the law books' energy went into controlling exactly whom they married.
- NOT claimed: a mechanism for the sex bias. Violence, coercion and reproductive advantage are indistinguishable in a genome. What the genome shows is that local male lines lost.
- Method: mtDNA traces one unbroken mother-to-daughter line, as the Y traces one father-to-son line. Neither is a genome-wide ancestry estimate, and nothing here converts one into the other.
- The R-37 reading is an inference from skeletal affinity in a single cemetery and is provisional.
Observation, model, inference — which is which
the asymmetry is measured; what it means is not
The page draws social conclusions from a genetic asymmetry. Separating the measurement from the reading is exactly what a visitor needs, and the register already holds that distinction per claim.