Made visible · one expansion, four outcomes

The Hittites spoke Indo-European. Their DNA carries no steppe marker.

The steppe expansion is usually told as one wave breaking outward from the Pontic-Caspian grasslands, carrying Indo-European languages wherever it reached. The ancient DNA tells something stranger. It struck Europe hard around 3200 BCE. It reached South Asia a full millennium later. It left only a fraction of Mycenaean Greece. And in Bronze Age Anatolia — where Indo-European is attested in writing — no Yamnaya-derived steppe ancestry has been detected in the genomes sampled so far. Whatever happened, it did not happen the same way twice.

The Anatolian zero: better supported, and not quite a zeroBetter supported: this page rests on Damgaard et al. 2018, 74 genomes. Lazaridis et al. 2022 sequenced 727 ancient individuals across Anatolia and its neighbours over 10,000 years and found Anatolia “remarkable for its lack of steppe ancestry down to the Bronze Age”. The page’s most striking negative is one of the best-sampled results in the field. And not quite a zero: newer work reports that the Caucasus–lower Volga population — the source population from which the Yamnaya themselves descend — contributed roughly one-tenth of the ancestry of Bronze Age central Anatolians, the Hittite speakers. That is not Yamnaya-derived steppe ancestry, and it does not restore the simple migration story. But “no steppe ancestry” and “no ancestry from the population the steppe pastoralists came from” are different statements, and only the narrower one is supported. Both papers have been reached at secondary level here; neither has been opened, and the Caucasus–lower Volga figure is recorded as needing verification against its source.

Four regions, four different answers.

the comparison

Region	Steppe ancestry	Timing	Source
Europe	Massive	3200–2500 BCE	Chintalapati & Moorjani 2022
South Asia	Yes — c. 1,000 years later	1800–1500 BCE	Narasimhan; Damgaard
Anatolia (Hittite)	No Yamnaya-derived steppe ancestry; ~10% Caucasus–lower Volga	IE attested regardless	Damgaard 2018; Lazaridis 2022 (727 individuals)
Minoan Crete	None	¾ Anatolian farmer	Lazaridis 2017
Mycenaean Greece	Only ~13–18%	source ambiguous	Lazaridis 2017

Two of these are negatives, and negatives are the most interesting results here. The Minoans — the first literate civilisation of Europe — derive about three quarters of their ancestry from the first Neolithic farmers of western Anatolia and the Aegean, with the remainder related to Caucasus and Iranian populations. No steppe component. The Mycenaeans differ, but only by roughly 13–18%, and even that is ambiguous between a steppe source and an Armenian one.

Two of the five results are zeroes. In prehistory, a well-sampled absence is often worth more than a presence.

What a zero does and does not mean.

detection limits, not proof of absence

This is where the argument has to be made carefully, because the obvious reading is wrong. "No steppe ancestry detected in Bronze Age Anatolia" is not the same as "nobody came." At least three explanations fit that result, and they are very different from one another:

The group was small. An intrusive elite — a warrior aristocracy, a ruling lineage, a military caste — can impose a language while contributing so little to the gene pool that it falls below the detection threshold of the individuals sampled. A tiny incoming minority is exactly the kind of event that leaves a language and almost no DNA.
The sampling missed them. Ancient DNA comes from whichever bodies happen to be excavated, preserved and published. The relevant people, period, region or social stratum may simply not be in the sample. Absence of evidence in a thin sample is weak evidence of absence.
Genuine cultural transmission. Language can spread through prestige, trade, administration or elite emulation with negligible gene flow.

Nobody ever claimed language spread requires replacement

It is worth stating plainly, because the point is often set up as a strawman: no serious account holds that Indo-European spread required population replacement. Languages travel by conquest, prestige, administration and trade as readily as by demography. So the Anatolian zero does not refute the steppe hypothesis for Anatolia — it constrains the mechanism. It says: if speakers came, they were few, or we have not found them.

What this does to the South Asian case.

the argument, stated precisely

South Asia is the region this platform documents, so the distinction matters here more than anywhere. The claim being made about South Asia has never been "the population was replaced." It is narrower and better evidenced: a detectable ancestry component arrives on a datable timeline, and it arrives alongside a language and a technology.

The timing is specific: Steppe_MLBA groups form around 1800–1500 BCE and are the vector by which steppe-related ancestry enters northern South Asia — a full millennium after the same ancestry reached Europe.
The package travels together: in Asia, unlike Europe, Indo-Iranian languages, chariots and horses spread together following the Sintashta culture. This is precisely the world the Rigveda's war hymns describe — the forts, the chariots, the named enemies.
It is unevenly distributed today: some named groups carry markedly higher steppe-related ancestry than others in the same region — the strongest signals of excess steppe ancestry in the source data are in specific jātis, Brahmin_Tiwari and Bhumihar_Bihar, not in “Brahmin” as a category; tribal groups carry very high AASI.

Tested at jāti level, not at the level of an aggregated category“Brahmin castes carry higher steppe and Iranian-related ancestry than non-Brahmin castes in the same region” is an aggregated-category claim. The evidence is jāti-level and uneven. In the source data the strongest excess steppe signals are in two specifically named groups; the highest modern steppe fractions (above 10%) appear in Punjabis and particular Brahmin samples; reported ranges differ by region, with northern upper-caste groups higher than southern Brahmin groups. A published critique notes that the study does not explain why those particular jātis peak. This matters for the museum’s own rule that varṇa is a classification and jāti is the endogamous unit: a statement about “Brahmin castes” pools dozens of jātis that have not intermarried for two millennia. “Ancestry proportions track the social hierarchy” is the strongest version of the claim and the least supported — what the data shows is a correlation with some groups at the top, not a proportional gradient down a hierarchy — and it is not made here. Narasimhan et al. 2019 is known to this museum at snippet level.

The Anatolian result does not weaken this. If anything it clarifies it: the steppe expansion produced a strong genetic signal in South Asia, which is more than it managed in Anatolia or Crete. Where the signal is present and datable, it can be reasoned from. Where it is absent, the honest conclusion is that we cannot tell from genetics alone.

The horse story is wrong for Europe and right for Asia.

273 ancient horse genomes

One of the oldest elements of the standard account has been overturned. The picture of mounted Yamnaya warriors riding into Europe is explicitly rejected by the largest study of ancient horse genomes yet conducted: the modern domestic horse lineage spread across Eurasia only from about 2000 BCE, long after the Yamnaya expansion into Europe around 3000 BCE. Horses in Corded Ware assemblages are limited, and the study concludes that horses were not the primary driver of Indo-European's initial spread in Europe.

But the same paper finds the opposite for Asia. There, the spread of domestic horses in the early-to-mid second millennium BCE was concurrent with the spread of chariotry and Indo-Iranian languages, whose earliest speakers are linked to populations preceding Sintashta. The chariot-warrior package — horse, spoked wheel, and the mythology that goes with them — is specifically the Indo-Iranian and South Asian story, not the European one.

Librado, P., Khan, N., Fages, A., et al. (2021). "The origins and spread of domestic horses from the Western Eurasian steppes." Nature 598(7882): 634–640.

DOI10.1038/s41586-021-04018-9

TypeSecondary — peer-reviewed; 273 ancient horse genomes

SupportsThat modern domestic horses expanded from c. 2000 BCE, rejecting horseback riding as the driver of the Yamnaya expansion into Europe; and that in Asia horses, chariotry and Indo-Iranian languages spread together after Sintashta.

Does not supportAny claim about who spoke what. This is horse genomics; the language association is inferred from the archaeological correlation, not demonstrated genetically.

Who is telling us this.

concentration risk

Three citations, one laboratory

Lazaridis, Narasimhan, Patterson, Moorjani and Chintalapati are all of the David Reich Laboratory (Harvard Medical School / Broad Institute) or its lineage — a lab that had, as of 2020, produced the majority of all published ancient-DNA data in the world. Citing them together is one evidence stream cited several times, not several independent confirmations. Such citations are routinely presented as mutual corroboration; they are not.

The independent check on this page is Damgaard et al. 2018 (Copenhagen), a separate teamThe concentration flag, and the paper added to itThis section names five authors, identifies them as one laboratory and its lineage, and states that citing them together is one evidence stream cited several times. Lazaridis et al. 2022, above, is also Reich-lab work. It makes the Anatolian zero far better sampled; it does not make it independently confirmed. Damgaard remains the only independent check on this page. whose findings partly contrast with the dominant framing: no steppe ancestry in Bronze Age Anatolia, West Eurasian ancestry entering South Asia before and after but not during the Yamnaya expansion, and Yamnaya-related migrations having "limited direct genetic impact in Asia." A contrasting result from an independent team is worth more than three concordant ones from the same source. See the Ledger.

Damgaard, P. de B., Martiniano, R., Kamm, J., et al. (2018). "The first horse herders and the impact of early Bronze Age steppe expansions into Asia." Science 360(6396).

DOI10.1126/science.aar7711

TypeSecondary — peer-reviewed; 74 ancient whole genomes; independent of the Reich lab

SupportsNo steppe ancestry detected in Bronze Age Anatolia when Indo-European is attested there; West Eurasian ancestry entering South Asia before and after, but not during, the Yamnaya expansion; limited direct Yamnaya genetic impact in Asia.

Does not supportThat no Indo-European speakers entered Anatolia. 74 genomes across a vast region cannot exclude a small or unsampled group. This is a detection limit, not a demonstrated absence.

What this page refuses.

on both flanks

That steppe ancestry equals Indo-European speech. Anatolia shows the two can come apart. Ancestry is evidence about people, not directly about language.
That a genetic zero proves nobody came. Small elites and thin sampling both produce zeroes. Stated as a limit, not a finding.
That Egypt or Phoenicia seeded the Aegean. Lazaridis finds no measurable Levantine or African influence in Minoans or Mycenaeans — a clean negative result.
Out-of-India. The documented direction of Steppe_MLBA gene flow is south and east; no source examined here supports a reversal.
That ancestry proportions imply cultural worth or priority. They do not. Ancestry is not achievement, and the genetics of caste is evidence of a mechanism, not of any group's nature.

Mela Keela is an independent digital museum. Sources, evidentiary limits and review status are identified on individual pages.