Cousin marriage, polyandry and matriliny in South Asia — and why the north–south story does not survive contact with Iran.

Zoroastrian tradition prescribes xwēdōdah — next-of-kin marriage. It is discussed in the Avestan and extensively in the Pahlavi literature, where it is not tolerated but praised: a meritorious act, among the most efficacious a person can perform. The relationships named include brother and sister, father and daughter, mother and son.

That is closer than cross-cousin marriage. It is closer than uncle–niece marriage. It is the closest degree any documented system has endorsed.

The Iranian branch of Indo-Iranian endorsed sibling marriage while the Indic branch was computing seven generations of prohibition.

So "Indo-European kinship is exogamous" cannot be maintained. The two branches of Indo-Iranian — which share a proto-language, a pantheon, a ritual vocabulary and a priestly class — sit at opposite ends of the entire human range on this question.

Which means the northern Indian exogamy rule is not an inheritance. It is a development, and it developed after the split — consistent with this platform's own count: sapiṇḍa and sagotra occur zero times across 660,000 words of Vedic text.

1 · FBD and MBD are genetically identical. Both are first cousins; both give a coefficient of relationship of 0.125 and an offspring inbreeding coefficient of 0.0625. The difference between them is entirely social — which lineage the spouse comes from — and has no genetic consequence whatever. Any chart that separates them by "closeness" is wrong.

2 · Elamite royal sibling marriage is the same degree as Egyptian, and Mesopotamian cousin marriage is the same degree as Dravidian cross-cousin marriage. The first version placed them differently on impression rather than criterion.

3 · Genesis is not one practice. Abraham and Sarah are half-siblings; Jacob marries MBD; Isaac marries a patrilateral kinswoman. Those are three different degrees and they now appear as separate rows. Comparing "Genesis" to "Dravidian" as single blocks was the error — the correct comparison is Jacob's marriage to Dravidian cross-cousin marriage, and they are the same thing.

The two outliers on this chart are Zoroastrian Iran at one end and Dharmaśāstra north India at the other. Both are Indo-Iranian. Cross-cousin marriage is unremarkable by comparison — it is the same degree Athens required of an heiress and the same degree Jacob married at.

The unusual practices in this comparison are Iranian sibling marriage and north Indian seven-generation exogamy. Cross-cousin marriage is the ordinary human middle.

Sharing a Y or mitochondrial haplogroup carries no health consequence on its own. The Y chromosome has roughly fifty functional genes; mitochondrial DNA has thirty-seven. Neither recombines. Millions of unrelated people share a haplogroup, and haplogroup identity says almost nothing about how much DNA two people share overall.

The health risk in close-kin marriage is autosomal. It comes from identical segments inherited on both sides across the twenty-two ordinary chromosome pairs — recessive variants meeting themselves. That is what the inbreeding coefficient measures, and it is 0.0625 for any first cousin, cross or parallel.

FBD marriage concentrates one Y-lineage generation after generation. Cross-cousin marriage does not. The disease risk is the same; the population-genetic signature is not.

So the two systems are indistinguishable by inbreeding coefficient and distinguishable by Y-chromosome diversity. A population with generations of FBD marriage should show reduced Y-haplogroup variety relative to its autosomal diversity; a cross-cousin-marrying population should not. That is a testable difference, and it is the right way to separate the two systems in genetic data — better than the consanguinity percentages, which cannot tell them apart at all.

The Rigveda contains a fourteen-verse dialogue in which a woman asks her twin brother to sleep with her, and he refuses. It is not a passing reference; it is a sustained argument with both sides given.

Her case: "Even in the womb Tvaṣṭar, shaping all forms, made us consorts" (v.5) — they were created as a pair. And: "I am possessed by love of Yama, that I may rest on the same couch beside him" (v.7).

His case: "Shall we do now what we never did before? We who spoke rightly, now talk impurely?" (v.4). "Great is the law of Varuṇa and Mitra" (v.6) — he invokes the vow-system, the same vrata order this platform finds attached to the Ādityas.

And the decisive line, v.12: "I will not fold my arms about your body: they call it sin when one comes near his sister."

Not "it is forbidden." Not "the law says." They call it sin — an appeal to what people say, offered by a god, in the oldest text in the corpus.

And v.10 is stranger still: "There will come succeeding times when brothers and sisters will do acts unmeet for kinsfolk." The text predicts the practice rather than ruling it out.

She gets the last word against him: "Alas, you are a weakling, Yama; we find in you no trace of heart or spirit" (v.13).

So the Vedic position is not a prohibition. It is a debate, with the woman arguing from creation and the man from convention, and the man's refusal resting on what people call things rather than on any stated rule. Placed beside the Prajāpati passage — where the act occurs and the Rigveda supplies no punishment — the earliest layer of the tradition is discussing close-kin union, not legislating against it.

The question is why a system with extreme caste endogamy also needs extreme lineage exogamy. They look contradictory. They are not.

That combination does something neither rule does alone: it makes the jāti the reproductive unit.

Without lineage exogamy, families would marry internally and the caste would fragment into self-sufficient lineages. Marriage would build no ties across it.

Without caste endogamy, exogamy would push marriage across caste lines and dissolve the boundary entirely.

Exogamy forces alliances outward; endogamy caps how far outward they can go. The caste is what is left in between.

So the seven-generation rule is not in tension with endogamy — it is the instrument that converts a set of lineages into a single bounded population. Every marriage is forced to cross a family and forbidden to cross a caste, and the cumulative effect over generations is a network of obligation whose outer edge is exactly the caste boundary.

And the chronology fits. Sapiṇḍa and sagotra: zero across 660,000 words of Vedic text. Jāti: zero across the same corpus. Both terms arrive with the Dharmasūtras, in the same window. They are not two developments that happened to coincide; they are one mechanism with two halves.

When Draupadī is to marry all five brothers, the text does not pass over it. Drupada objects. Yudhiṣṭhira has to argue the case. Vyāsa is brought in to settle it, and settles it by supplying a divine explanation.

A practice needing that much justification inside the text is a practice the text's compilers were uncomfortable with — and could not remove. It was too central to the story to cut. The same pattern this platform documents elsewhere: what could not be deleted was explained.

And the Himalayan cases matter for the north–south framing. Fraternal polyandry is documented among Indo-Aryan-speaking communities in Jaunsar-Bawar and Kinnaur, tied to the same problem it solves in Ladakh: keeping a small mountain holding from being divided between brothers.

Marriage systems track land and terrain at least as closely as they track language.

Only the eldest Nambūdiri son married a Nambūdiri woman. Younger sons formed sambandham unions with Nayar women. Those children were Nayar, belonged to the mother's taravāḍu, and inherited nothing from their father.

Why it existed in Kerala and nowhere else is answerable, and the answer is structural.

The arrangement requires a matrilineal community to receive the children. A younger son's offspring have to belong somewhere — and only a matrilineal system can absorb them without a claim on the father's estate, because in a matrilineal system the father was never the source of a child's belonging in the first place.

Under patrilineal descent the same arrangement is impossible. The children would be nobody's: not their father's heirs, because the estate is undivided, and not their mother's, because a patrilineal lineage does not transmit through her. The structure has nowhere to put them.

Elsewhere, brahmin families facing the same problem — too many sons, one estate — used other solutions: younger sons entered temple service as salaried priests, or took monastic ordination, or simply remained unmarried dependants inside a joint household, fed by the estate but not dividing it.

All of those keep the younger son inside the community and childless. The Kerala solution is the only one that gives him children, and it can do that only because there was a neighbouring system that would claim them. Remove the matrilineal taravāḍu and the arrangement has no destination for the offspring, and collapses back into one of the other three.

The arrangement did not create matriliny and did not oppose it. It was built on top of it, and could not have been built anywhere else.

Whether it was designed as a response to matriliny or co-evolved with it is not settled. What can be stated is the structure: a landholding rule that kept brahmin property undivided, met the reproductive needs of unmarried younger sons, and placed the cost of raising the resulting children on a different community's estate.

The system did not survive contact with codified law. Once sambandham was ruled not to be marriage and the taravāḍu was made partible, both halves of the arrangement collapsed — though not symmetrically, and the matrilineal side lost more.

When parents are related, their child inherits identical stretches of DNA from both sides. These are runs of homozygosity, and their total length gives the inbreeding coefficient F.

Reported rates of consanguineous marriage are among the highest in the world in South India — figures in the range of a fifth to nearly half of unions in parts of Tamil Nadu, Andhra Pradesh and Karnataka, depending on community and survey. In north India the comparable figures are a few per cent.

This shows up directly in genomic data as elevated ROH in southern populations. It is one of the strongest north–south genetic gradients in the subcontinent, and unlike ancestry proportions it is not about deep history at all.

Sapiṇḍa and sagotra occur zero times across roughly 660,000 words of Vedic text. The exogamy calculation that defines north Indian marriage — seven generations paternal, five maternal, plus the gotra bar — appears with the Dharmaśāstra, not before.

And gotra in the Rigveda does not mean lineage. Its nine occurrences carry the older sense: a cowpen, an enclosure — which is why Indra is said to break open the gotra to release the cattle. The word for the unit that governs who you may marry began as the word for a cattle stockade.

The rule that defines northern marriage is younger than the texts it claims authority from.

Muslim communities across South Asia practise cousin marriage — but predominantly father's brother's daughter (FBD). That is the Middle Eastern and North African pattern, and it is the opposite of the Dravidian rule: in the Dravidian system FBD is a sister and marriage to her is incest.

So consanguinity figures for Gujarat and Maharashtra may be counting two entirely different systems together. An elevated rate in a Muslim community there is evidence of Islamic marriage preference, not of a Dravidian substrate — and a survey that reports only "consanguinity %" cannot distinguish them.

Any west-coast substrate argument has to separate FBD from cross-cousin marriage first. The two are not variants of one practice; they encode incompatible rules about who counts as a sibling.

Nahali (Nihali) is spoken by a small community in Madhya Pradesh and Maharashtra. Its vocabulary is layered: Korku (Munda) loans, Dravidian loans, Indo-Aryan loans — and a core that has no accepted classification at all.

Kuiper argued that this unclassified core is a genuine relic — a survival from a language family that predates Munda, Dravidian and Indo-Aryan alike in central India. If that is right, Nahali is the only living trace of a fourth linguistic layer, and it is spoken by a few thousand people.

Whether Parpola specifically classified it as Austroasiatic, this page does not assert without checking the passage. The classification is genuinely disputed: some treat it as Munda with heavy borrowing, others as an isolate. What is not disputed is that a substantial part of its core vocabulary has no source anyone has identified.