Evidence page · the present tense
The genome dates the sealing to two thousand years ago. The survey data says it has not opened.
Everything else on this platform about endogamy is archaeology: founder events, bottlenecks, the moment mixture between groups became rare. This page is the same phenomenon measured by household survey in living memory. The largest non-government pan-Indian survey put the inter-caste marriage rate at under six per cent — with no upward trend across four decades of marriages. The practice that shows up in the DNA as a two-thousand-year-old event is also a fact about weddings held last year.
5.82%
inter-caste marriages, IHDS-II
~42,000
households surveyed
1951–2012
marriage years covered
no trend
across four decades
The flat line.
what four decades of data show
By state, and how far apart they are.
the same country
| State | Married within caste | Note |
|---|---|---|
| Madhya Pradesh | ~99% | Inter-caste marriage under 1%. |
| Himachal Pradesh | ~98% | |
| Chhattisgarh | ~98% | |
| Gujarat, Bihar | ~89% | Highest inter-caste rates recorded, above 11%. |
Even the most open state in the survey married within caste in nearly nine cases out of ten. Inter-caste marriage is reported as lowest in less developed villages and in the regions where caste councils are strongest, and higher in the northeast, where the caste order was never as fully established.
Even crossing the line is usually arranged.
the finding that reframes the rest
sort by rate; filter to a state
| Measure | Finding |
|---|---|
| 73% | of marriages reported as arranged by parents |
| ~70% | of women said they met their husband on the wedding day or at the gauna |
| 63% | of inter-caste marriages were also arranged by parents or relatives |
| 98% | of inter-caste couples lived with parents immediately after marriage |
| 1 in 4 | respondents knew anyone in their community who had married across caste |
A flat national line hides enormous state variation. Sorting reveals it; the static table does not.
The intuitive picture of an inter-caste marriage is a couple defying their families. In the data, most of them are arranged by those families. Which means the small figure is not a measure of how many young people rebel — it is a measure of how often parents choose to cross, and they almost never do.
And that in turn tells you where the boundary is maintained. Not primarily in individual prejudice, but in the marriage decisions of households — the same unit that appears in the genetic record as a lineage.
What enforces it.
and the word will be familiar
Caste councils — khap panchayats among the Jats of Haryana, Punjab, western Uttar Pradesh and parts of Rajasthan — issue rulings on marriages that violate community norms, and have operated with official tolerance for many years.
A khap ruled that the couple belonged to the same gotra — Banwāla — and were therefore to be regarded as brother and sister, making their marriage invalid and incestuous. They married anyway. They were abducted and killed by the woman's relatives. The state government expressed no concern about the khap's ruling. The subsequent prosecution was among the first in which khap members were convicted in such a case.
The unit invoked was gotra — the exogamous patrilineal descent group with a named apical ancestor. The same structure that appears in the genetic record as the mechanism behind the male-lineage bottleneck is the structure a village council cited in 2007 to declare a marriage incestuous between two unrelated people. It is not a survival. It is in continuous operational use.
And untouchability is reported directly.
by the households practising it
In the same survey, 30% of rural and 20% of urban households said they practise untouchability. That is self-report — households stating a practice that is constitutionally abolished and criminally punishable.
Self-reported figures for a prohibited practice are a floor, not an estimate. Whatever the real number is, it is not lower than this one.
How firm the numbers are.
a genuine methodological spread
| Source | Rate | Method |
|---|---|---|
| IHDS-II (2011–12) | 5.4–5.8% | Married women asked directly whether the husband's family is the same caste as their natal family. |
| NFHS (2005–06) | ~11% | Indirect — comparing the caste group each spouse reported separately. |
| NFHS-5 (2019–21) | 17% urban | Different instrument and a different sub-population. |
Caste self-identification is unstable. As the demographer who led IHDS-II put it, the same person will say "I am Baniya" one day and "I am Modh Banik" the next, and both are correct. Whether a marriage counts as inter-caste depends on which level of the hierarchy the question addresses — jāti, sub-jāti, or broad varṇa category — and the surveys do not all ask the same thing.
The spread is real and this page does not pretend otherwise. But the range runs from about 5% to about 17% in an urban sub-sample. At the most generous reading available, more than four marriages in five are still within caste, and the direct-question estimate shows no upward movement over forty years.
What this closes.
- The genetic and the contemporary record are describing one continuous thing. The founder events, the end of mixture around two thousand years ago, and a survey finding of 95% same-caste marriage in 2012 are not two subjects. They are the same practice at two points in its duration.
- Endogamy is not a residue. It did not survive modernity by inertia. It has been maintained through independence, land reform, urbanisation, mass education and half a century of legal permission, with no measurable loosening.
- And it is maintained by households, not by doctrine. Which is why changing religion does not end it, and why the law that abolished untouchability has not abolished the practice 20 to 30 per cent of households report.
The honest limits.
- Sources: IHDS-II (2011–12), conducted by NCAER with the University of Maryland — 42,152 households, 39,523 ever-married women, marriages from 1951 to 2012 — for the 5.4–5.8% rate, the state figures, the arrangement percentages and the untouchability self-reports; Ray, Chaudhuri and Chowdhury (2017/2020) for the 63% and 98% figures on inter-caste marriages; NFHS 2005–06 and NFHS-5 (2019–21) for the alternative estimates; contemporary reporting and the court record for the Manoj–Babli case.
- The chart is a schematic of a reported finding, not a plot of the underlying series, and is labelled as such.
- The estimates are method-dependent and the spread is set out above rather than resolved.
- IHDS-II is now over a decade old. A more recent nationally representative direct estimate on the same question does not appear to be published.
- NOT claimed: that ancient endogamy and present endogamy are the same institution unchanged. Practices persist by being continually remade; continuity of function is not identity of form.