The Timeline of Endogamy: How Geneticists Date the Rise of India's Caste Structure

A geneticist looking at a modern Indian genome can often guess, with some accuracy, which endogamous (marrying only within one's own defined group) community a person's ancestors belonged to — sometimes without ever seeing a name or a surname. That is not mysticism. It is arithmetic done on chromosomes.

The arithmetic in question involves counting long stretches of DNA that two people share unbroken, called identity-by-descent (IBD) segments. Every generation, recombination — the shuffling of DNA during reproduction — chops parental chromosomes into smaller and smaller pieces before passing them to children. So a long shared segment means a recent common ancestor; a short one means a distant one. If you measure the segment lengths shared between two people from the same community across thousands of genomes, you can estimate, in generations, how long ago their ancestors formed a closed mating pool.

This is the method behind one of the more startling findings in Indian population genetics: that many jati (endogamous sub-caste) groups have been marrying almost exclusively within themselves for somewhere between roughly 1,900 and 4,200 years, depending on the group Moorjani et al., 2013, American Journal of Human Genetics.

Two ancestries, one mixture event, then a wall

The foundational work here comes from the Reich lab's 2009 paper describing two major ancestral populations that contributed to almost all Indian genomes today: Ancestral North Indians (ANI), related to Central Asian and West Eurasian populations, and Ancestral South Indians (ASI), a lineage found nowhere outside South Asia Reich et al., 2009, Nature. Later work refined this into a more detailed model involving Ancient Ancestral South Indians (AASI), Iranian-related farmer ancestry, and steppe pastoralist ancestry arriving in overlapping waves Narasimhan et al., 2019, Science.

The 2013 Moorjani study, led by geneticist Priya Moorjani, then asked a narrower and more testable question: not "where did the ancestry come from," but "when did people stop mixing it further?" The team analyzed genome-wide data from 73 groups across the Indian subcontinent and modeled how IBD segment lengths decayed over generations within each group Moorjani et al., 2013. Their result: ANI-ASI mixture was widespread and ongoing until somewhere between about 1,900 and 4,200 years before present, after which many groups' admixture proportions essentially froze in place — the genetic signature of populations that stopped marrying outward.

Why a date range, not a date

The wide bracket is not sloppiness. It reflects real biological and statistical noise: mutation rates, generation-length assumptions, and the fact that different communities appear to have sealed themselves off at different times, not on a single calendar date. Narasimhan and colleagues' later, much larger dataset — combining ancient DNA from over 500 individuals across Central and South Asia with modern genomes — pushed the analysis further, showing that steppe-related ancestry arrived in South Asia around 2000–1500 BCE, filtering in over centuries rather than in one migration event, and that strict endogamy in many groups appears to postdate this mixing by centuries Narasimhan et al., 2019.

This matters because it separates two different questions that popular discussion often collapses into one. When did ancestral populations mix? is a different question from when did they stop mixing and start forming the endogamous units recognizable in the historical and ethnographic record? The genetic data says the mixing was gradual and the stopping was also gradual and staggered — not a single dramatic founding event, and not evidence for or against any particular textual account of how social stratification began.

What the method can and can't tell you

IBD-based dating has limits worth stating plainly. It estimates when a population's effective mixing stopped — meaning gene flow dropped below a detectable threshold — not necessarily the exact year a social rule was enacted or an inscription written. It also assumes a roughly constant generation time (commonly modeled around 25–29 years in these papers) and can be distorted by population bottlenecks, migration, or founder events unrelated to marriage rules. Scholars debate exactly how to translate a genetic "mixing stopped" date into a social history of when endogamy became normative practice versus when it hardened into rigid, policed boundaries.

What the data does support, cautiously, is a picture: ANI-ASI admixture happened over an extended period following the arrival of Iranian-related and steppe-related ancestries, and strict endogamy — genetically detectable as a sharp drop in cross-group IBD sharing — set in afterward, at different times for different groups, within roughly the last two to four thousand years. That is a genetic pattern, not a verdict on caste's moral history, and the two should not be confused.