6.2 — The Vedic Age and the Aryan Question, Honestly

The Sanskrit word for father is pitr. In Latin it is pater, in Greek pater, in Old English fæder, in Persian pidar. Sanskrit matr, Latin mater, Greek meter, English mother. Sanskrit trayas, Latin tres, Greek treis, English three.

In 1786 a British judge in Calcutta named William Jones, who had learned Sanskrit to read Hindu law in the original, told the Asiatic Society that Sanskrit bore to Greek and Latin "a stronger affinity, both in the roots of verbs and in the forms of grammar, than could possibly have been produced by accident", and that all three must have "sprung from some common source, which, perhaps, no longer exists."

He was right, and the discovery founded the science of historical linguistics. It also started an argument that is still running, and this chapter is the most careful in this Part because the argument has become politically charged in ways that make honest statement difficult and therefore necessary.

What is not in dispute

The Indo-European language family is real. Sanskrit, Persian, Greek, Latin, the Celtic, Germanic, Slavic and Baltic languages, Armenian and Albanian all descend from one ancestral language. This is established by the same kind of systematic correspondence that establishes any family relationship, and it is not seriously questioned by any linguist.

Sound changes are regular, which is what makes the method work. Latin p corresponds to English f systematically — pater/father, piscis/fish, pes/foot — and the rule was worked out in the nineteenth century. Regular correspondences of this kind cannot be produced by borrowing or coincidence.

In India, the northern languages are Indo-European and the southern ones are not. Hindi, Bengali, Marathi, Gujarati, Punjabi, Odia, Assamese and Sindhi descend from Sanskrit's relatives. Tamil, Telugu, Kannada and Malayalam belong to the Dravidian family, which is unrelated. Munda and Tibeto-Burman families are also present.

And Sanskrit's closest relative is Avestan, the language of the Zoroastrian scriptures of Iran. The two are so close that passages can sometimes be converted between them by regular sound rules. The Rigveda's deva is Avestan daeva; asura is ahura; soma is haoma; Mitra is Mithra. Strikingly, the two traditions reversed the moral valuation of two of these terms — deva is divine in India and demonic in Iran; asura is demonic in India and divine in Iran — which suggests a shared origin followed by a religious split.

The Rigveda itself

The Rigveda is the oldest text in any Indo-European language, and it is a remarkable document. Roughly 1,028 hymns, composed in an archaic Sanskrit, transmitted orally with the extraordinary fidelity described in Chapter 3.11, and generally dated on linguistic grounds to roughly 1500–1200 BCE, with the older layers possibly earlier.

What it describes: a pastoral society whose wealth is cattle, organised in tribes led by a rajan with an assembly, fighting each other and a people called the dasa or dasyu, using horse-drawn chariots with spoked wheels, performing fire sacrifices, and drinking a pressed plant juice called soma.

What it does not describe: cities, brick architecture, sea trade, or a script. It is not a description of the Indus civilisation, and this is one of the strongest arguments in the whole debate. The Indus cities had baked-brick urbanism, standardised weights, drainage and seals; the Rigveda's world has none of it.

Its geography is the northwest. The rivers it names most are the Indus and its tributaries and the Sarasvati; the Ganges is mentioned only once or twice, in late hymns. Later Vedic texts move steadily eastward, and the whole corpus reads as a society shifting from the Punjab into the Ganges-Yamuna region over centuries.

What linguistics established

The reconstructed ancestral language, Proto-Indo-European, has vocabulary that points to a specific environment.

It has words for horse, wheel, axle, yoke, and wagon. Since the wheeled vehicle is not attested anywhere before about 3500 BCE, the parent language cannot be much older than that — otherwise the daughter languages would not share inherited words for it.

It has words for wool, honey, bee, birch, and cold-climate animals, and words for domesticated cattle and sheep. It lacks inherited words for tropical animals and plants — no shared word for tiger, elephant, rice, or bamboo. Where daughter languages have such words, they are borrowings.

These vocabulary constraints point to a temperate, pastoral, horse-and-wagon society, and the leading candidate on linguistic grounds has long been the Pontic-Caspian steppe, north of the Black and Caspian Seas, around 4000–3000 BCE. This is the Steppe Hypothesis, associated with Marija Gimbutas and later David Anthony.

The main rival, the Anatolian Hypothesis of Colin Renfrew, put the origin in Anatolia around 7000 BCE, spreading with farming. It is now generally regarded as failing the wheel-and-horse vocabulary test — the language would be far too old for the shared wagon words.

What ancient DNA showed

This is where the last decade changed the picture, and the results are worth stating precisely because they are frequently misreported in both directions.

Method. DNA is now recovered from ancient skeletons across Eurasia, and populations can be modelled as mixtures of earlier populations. Hundreds of ancient genomes from the steppe, Iran, Central Asia and South Asia have been published since 2015.

Result 1: a large steppe expansion happened. Populations from the Pontic-Caspian steppe, associated with the Yamnaya culture, expanded massively westward into Europe from around 3000 BCE — in some regions replacing a large fraction of the existing male lineages — and eastward into Central Asia.

Result 2: steppe ancestry reached South Asia, and it arrived in the second millennium BCE. Modern South Asian populations carry a component that models as steppe-derived, and ancient samples from Central Asia and the northern subcontinent show it appearing in the relevant window.

Result 3: the Indus population did not have it. The single genome recovered from Rakhigarhi, published in 2019, models as a mixture of Iranian-related farmer-related ancestry and South Asian hunter-gatherer ancestry, with no detectable steppe component — consistent with a larger set of individuals from Indus-related contexts found at a site in Iran.

Result 4: the Iranian-related component in South Asia separated early, before the spread of west Asian farming, which means it does not indicate a migration of Fertile Crescent farmers into India. Agriculture in South Asia has substantial local development (Chapter 3.7), and the genetics supports that.

Result 5: steppe ancestry is distributed unevenly today — higher in the northwest, lower in the south and east, and, within regions, correlated with traditionally higher-status and priestly groups in several studies. This is a documented statistical pattern and it is uncomfortable, and this book states it rather than omitting it.

What this does and does not mean

Here is where care is essential.

It does not mean an invasion. The nineteenth-century picture of chariot-borne Aryans destroying Indus cities is dead, and Chapter 4.3 gave the archaeological reasons. There is no destruction layer, no evidence of conquest, and the Indus decline precedes the relevant period by centuries.

What the evidence supports is migration — probably in waves, over generations, of pastoralist groups moving through Central Asia into the northwest — into a region whose urban civilisation had already de-urbanised for climatic and hydrological reasons. Incoming groups mixed with a much larger existing population. The overwhelming majority of Indian ancestry is indigenous; the steppe component is a minority contribution everywhere, though not a trivial one.

It does not mean anyone is foreign. A migration in 1500 BCE has no bearing on anyone's claim to be Indian in 2026, any more than the Anglo-Saxon migration makes the English foreign in England. Every population on Earth is a mixture of earlier movements, and Chapter 3.2 established that as a general fact about our species.

And it does not settle what the Indus people spoke. The script is undeciphered (Chapter 3.11 and 4.3). Dravidian is the leading guess — supported by the isolated Dravidian language Brahui in Balochistan, and by apparent Dravidian and Munda substrate features in Sanskrit — and it remains a guess.

The Out-of-India position, assessed fairly

The alternative position — that Indo-European languages originated in India and spread outward — is held by some Indian scholars and has a substantial popular following. It deserves to be assessed rather than dismissed, and this book will do that.

The arguments made for it: continuity in the archaeological record without a clear intrusive horizon; the Rigveda's own geography being firmly Indian; the identification of the Sarasvati with the Ghaggar-Hakra, whose drying is dated to before the conventional Vedic period, which would push the Rigveda earlier; and the claim that the migration model is a colonial construct designed to delegitimise Indian civilisation.

The points that are fair. The colonial-era framing of "Aryan invasion" was genuinely tied to racial theories, was used politically, and deserved to be discarded — and it has been. The Rigveda is indeed geographically Indian. The Sarasvati question is genuine and its dating is argued.

Where the position fails, on the evidence. The linguistic problem is decisive: if Indo-European originated in India, the reconstructed parent language should show a South Asian environment, and it does not — no inherited words for tropical flora and fauna, and inherited words for temperate ones. The genetic problem is that the ancestry flow is directional: steppe ancestry appears in South Asia after it exists on the steppe, and Indian-specific ancestry does not appear in ancient European samples in the way the reverse model requires. And Indo-European languages are older in written attestation outside India — Hittite in Anatolia and Mycenaean Greek are both attested before any Indian text was written.

The honest position: the migration model is supported by three independent lines of evidence — linguistics, archaeology and ancient DNA — that were developed separately and agree. The Out-of-India model has to explain away all three, and it has not done so. That is a scholarly judgement about evidence, and it carries no implication whatsoever about anybody's belonging or worth.

Vedic society, from the texts

What the texts show, with the caveat that they were composed by and for a priestly elite and describe an ideal at least as much as a reality.

Early Vedic society was pastoral, mobile and tribal. Wealth was cattle; the word for war, gavishti, literally means the search for cows. Leadership was a chief with a warrior assembly, not a bureaucratic king.

Religion centred on fire sacrifice. Offerings poured into fire and carried by Agni to the gods — Indra the storm and war god, Varuna the upholder of cosmic order, Agni, Surya, Soma. There were no temples and no images. The ritual was the religion, and the priests who knew it exactly held the power.

Later Vedic society, from roughly 1000 BCE, changed substantially as the population moved east into the Ganges valley, cleared forest with iron tools (Chapter 3.12), and settled into agriculture. Kingship became territorial and hereditary, rituals became enormous and expensive, and the priestly class became more formally dominant.

Varna appears in this period and its early form is not the caste system of later India. The word varna means colour or category, and the fourfold scheme — brahmana, kshatriya, vaishya, shudra — appears once in the Rigveda, in the Purusha Sukta, which is generally regarded as one of the latest hymns.

Three distinctions matter and are usually collapsed. Varna is the fourfold theoretical scheme. Jati is the several-thousand actual endogamous birth groups that structure real social life, which are regional, occupational and far more numerous. Untouchability, the exclusion of some groups from contact entirely, is a later development and is not present in the Rigveda.

And the evidence shows the early system was more fluid than the later one. Vedic texts contain examples of movement between categories, and there are sages of disputed or low origin in the tradition itself. Genetic studies indicate that endogamy — marrying strictly within group — became widespread only in the first millennium CE, with a fairly sharp transition. So the rigid, hereditary, endogamous system is roughly two thousand years old rather than four thousand, which is neither as ancient as its defenders claim nor as recent as some critics assert. Chapter 6.6 takes it up where the evidence for hardening is strongest.

Where this shows up in your life

Your language. If you speak an Indian language, you speak either a descendant of this Vedic-era Sanskrit family or a Dravidian, Munda or Tibeto-Burman language with centuries of Sanskrit borrowing. The words are in your mouth.

Your rituals. The fire sacrifice survives in the wedding ceremony's homa, in the namaste gesture's antiquity, in the sacred thread, and in mantras still recited in a language nobody has spoken natively for two thousand years — preserved by the memorisation techniques of Chapter 3.11.

And a habit of mind worth taking from this chapter. This is the most politically loaded question in Indian history, and there are strong incentives on all sides to state a conclusion more confidently than the evidence allows. The method that works is the one used here: separate the linguistic evidence from the archaeological from the genetic, ask what each independently supports, note where they agree, and say plainly where they do not. Chapter 16.4 turns that method into a general tool.

What the next page covers

By around 600 BCE the eastern Ganges plain had iron, forest clearance, surplus agriculture, towns, coinage and sixteen substantial states competing with each other. It also had an intellectual ferment in which the authority of the Vedic priesthood was being questioned from every direction. Chapter 6.3 covers that world — the sixteen kingdoms, the rise of Magadha, the second urbanisation, the invention of coinage in India, and the two men who walked out of it and changed Asia: Siddhartha Gautama and Vardhamana Mahavira.