In early 2025, a team out of Harvard and the Max Planck Institute published the genomes of four hundred and thirty-five people who had been dead for five thousand years, pulled from burial mounds scattered between the Volga and the Caucasus. The paper, led by Iosif Lazaridis and David Reich’s lab, proposed a new ancestor population for the entire Indo-European family - the Caucasus-Lower Volga cline, CLV for short, a group older and more obscure than the Yamnaya herders who had held the honor of “Proto-Indo-European homeland” for the better part of a century. Within weeks the paper had fractured its own field. Paul Heggarty, a linguist who has spent two decades trying to make archaeology and philology agree with each other, wrote that the study’s actual data amounted to a retreat from the steppe hypothesis dressed up as its confirmation. Ukrainian researchers refused to co-author with their Russian counterparts, splitting the work across two publications instead of one. And buried in the supplementary material was an inconvenient fact nobody quite knew what to do with: ancient Iranian populations show shockingly little steppe ancestry at all, for a culture that supposedly received its Indo-European language from the steppe.
This is worth sitting with, because it is not a footnote to the Aryan question. It is the Aryan question, still unresolved after a hundred and fifty years and now several hundred sequenced genomes, splintering along the exact fault lines - national, political, institutional - that Koenraad Elst identified decades ago as the actual architecture of the debate. The steppe theory was never simply data working itself out. It was a colonial-era racial category, “Aryan,” retrofitted first into a philological category and then, when that grew embarrassing, into a genetic one, each time promising the finality the previous version had failed to deliver. It promises finality again now. It will not deliver it now either, and the reason is instructive.
Consider what the CLV paper actually claims to have found: not a point of origin so clean it ends debate, but a genetic cline - a smear across geography and time, admixing with earlier populations, moving in directions the data cannot fully specify, producing daughter groups whose relationship to the parent stock the paper’s own authors argue about in the discussion section. This is what five thousand years of population movement actually looks like when you try to read it off of bones: gradients, not arrows. Ancestry components, not migrations with dates stamped on them. The confident diagrams that end up in textbooks - steppe pastoralists riding into India around 1500 BCE with horse, chariot, and language, ending the peaceful urban world of the Harappans - are not what the genetic evidence shows. They are what gets layered onto the genetic evidence by people who arrived at the question already believing that Vedic civilization must have come from somewhere else.
The Sindhu-Sarasvati civilization did not need this story assembled for it. It had already told its own story, in the one archive nobody thinks to check first because it does not come out of the ground: the Rgveda itself, composed over generations and preserved with a fidelity that has no parallel in the ancient world, transmitted orally with such precision that variant readings across regional recensions separated by centuries remain vanishingly rare. Shrikant Talageri did what should have been done a century earlier and read the text’s internal chronology as evidence in its own right, not as scripture to be mined for anthropological color. The oldest books of the Rgveda - six, three, and seven, in the order of their composition - name the rivers of the Ganga-Yamuna region and the tribes who lived among them. The later books, four, two, then five and the family books that follow, extend westward, naming the Sindhu, and only in the latest strata the rivers of Afghanistan and beyond. The direction embedded in the text’s own compositional layers runs east to west. An invasion from the northwest would have to write its own epic backward to produce this pattern, and no one has explained why a text supposedly composed by triumphant newcomers would begin by describing the interior of the country they had not yet reached.
Naditama - “most mighty of rivers” - is what the Rgveda calls the Sarasvati, and it names her more than any other river, hundreds of times, flowing pure from the mountains to the sea. Geology confirms that a river matching this description, the paleochannel now traced under the dry bed of the Ghaggar-Hakra, carried major flow until somewhere around 1900 BCE, after which its glacial and monsoon-fed sources were captured by the Sutlej and the Yamuna and it began to die. Michel Danino’s reconstruction of this hydrology in The Lost River is the definitive account, and its implication is simple enough that it should have ended the debate on its own: a text that describes the Sarasvati as a living, roaring, civilization-sustaining river cannot have been composed after the river stopped being any of those things. The AIT chronology needs the Rgveda composed around 1500 BCE, arriving with steppe migrants. The river makes that date impossible. You cannot sing hymns to a river’s flood in a century when the riverbed is already sand.
What actually happened along the Sarasvati after 1900 BCE was not an invasion. It was the largest population its basin could no longer support, moving east into the Ganga-Yamuna doab, carrying their gods, their meters, their fire altars, and their memory of the river with them - which is exactly the population movement the archaeological record shows: continuity of material culture, continuity of settlement pattern, no destruction layer, no burned cities, no replaced population, just people relocating along a hydrological gradient the way people have always relocated along hydrological gradients. This is, not incidentally, the same kind of movement CLV’s own authors are struggling to characterize honestly in their steppe data - not armies, not replacements, but clines, gradients, admixture happening over centuries in directions their model cannot cleanly arrow. The method that dissolves the steppe invasion into ambiguity is the same method that should dissolve the Aryan invasion of India into ambiguity, and for exactly the same evidentiary reason. It has not been applied evenly.
The horse and the chariot, invoked for a century as the unmistakable signature of the incoming Aryan, do not cooperate with the story either. Equid remains and imagery consistent with domesticated horses appear in Indian archaeological contexts earlier than the 1500 BCE arrival date requires, and the chariot itself - light-wheeled, spoked - is attested across a wide swath of Eurasia in a period too tangled and too contested to serve as anyone’s clean marker of ethnic arrival. B.B. Lal’s decades of fieldwork, from Hastinapura to Kalibangan, found no stratigraphic break where the invasion is supposed to have happened - only the steady, legible continuity of a civilization occupying the same land, adapting its material culture gradually, the way civilizations do when nobody has invaded them.
And when the Rgvedic peoples do appear outside India, they appear as people who have already left, not people who have not yet arrived. The Mitanni kingdom of northern Syria, flourishing around 1500 BCE, worshipped Mitra, Varuna, Indra, and the Nasatyas by name, used Sanskrit numerals to count the strides of chariot horses in a training manual, and gave their kings Sanskrit names two thousand kilometers from the Sindhu-Sarasvati basin. The standard account calls this a “cousin” relationship, an Indo-Iranian population that split off before reaching India and happened to retain similar gods. Talageri’s chronology makes a more economical claim: the Mitanni are a diaspora, a westward branch of a civilization already centered on the Sindhu-Sarasvati, carrying its religion outward the way Sanskrit later carried it outward again to Southeast Asia, except two and a half millennia earlier and with less institutional memory of the fact.
Subhash Kak and David Frawley have pushed the deep chronology further still, reading the astronomical references embedded in the Vedic corpus - the positions ascribed to the equinoxes, the seasonal markers encoded in ritual calendars - as pointing to composition dates well before 3000 BCE, with the later Vedanga Jyotisha’s precise calendrical calibration around 1300 BCE representing not the origin of Vedic astronomy but a late redaction of a much older astronomical tradition. This dating remains contested even among those sympathetic to it, and it should be held with appropriate care. But it points in one direction only, and that direction is backward, toward an antiquity the steppe model has no room for at all, not into an argument about whether Aryans arrived a little earlier or a little later than 1500 BCE, but into a civilizational timescale the entire question was never built to accommodate.
There is a reason none of this required a genome sequencer, and the reason is itihasa - “iti ha asa,” thus indeed it was - the Indian category for sacred-historical narrative, which the tradition never separated from cosmology the way modern historiography insists on separating memory from myth before either can be trusted. A civilization that keeps its own testimony, transmitted with the disciplined precision the Vedic reciters maintained for three thousand years before a single line was written down, is not a civilization waiting for outside verification. It already has an archive. The archive is oral, metrical, theological, and it does not apologize for being those things instead of being a bone. Western historiography, confronted with a textual tradition that predates and outlasts its own methods, has two options: read the text as evidence, the way it would read a cuneiform tablet or a Roman itinerary, or discount it as unreliable religious material requiring external correction. It has almost always chosen the second option for India specifically, for reasons Elst has traced back to a nineteenth-century need to keep Sanskrit civilization junior to the Bible, and reasons the Marxist historiography that dominated Indian universities after independence found equally useful for its own project of framing Hindu civilization as derivative, feudal, and foreign to the soil it occupies.
What makes 2025 interesting is not that the genetics have proven the Out-of-India position. They have not, and claiming otherwise would repeat the exact error being diagnosed here, just with the arrow reversed. What makes 2025 interesting is that the discipline built to settle this question in the invasion model’s favor keeps failing to settle it, keeps splintering into cline models and contested clines and authors who will not sign the same paper, keeps discovering populations like the ancient Iranians who have the language without the ancestry the model requires them to have. Every retreat, every hedge, every admission that the steppe data resists a clean directional story is a retreat from the confident nineteenth-century narrative that Indian civilization was written by outsiders. It is not an advance toward Out-of-India. It is the ground softening under a theory that was never as solid as its textbook diagrams pretended, in exactly the region where the text itself never wavered.
The individual holds memory the way a civilization holds itihasa - selectively, meaningfully, shaped by what needed to be remembered for the self to keep being itself. A person who loses that memory does not become free of their past. They become vulnerable to whoever offers to supply a replacement. This is worth naming plainly, because it is the actual stake in an argument that can otherwise look like a dispute between philologists: a civilization persuaded that its own memory is unreliable, that itihasa is myth requiring correction, that the river its oldest poetry sings about is a metaphor rather than a report, has ceded the authority to narrate its own origin to whoever currently controls the university department and the sequencing budget. It has ceded it repeatedly, and re-earned it repeatedly, most decisively in the last three decades in the work of Talageri, Danino, Kak, and Elst, none of whom needed a Harvard lab to tell them where the story starts.
Somewhere in the supplementary tables of the 2025 Nature paper is a cline that does not know it is describing, from ten thousand kilometers away and with considerably more expense, the same directional ambiguity that Talageri had already resolved from inside the text two decades earlier. The bones will keep arguing with each other for a while yet, splitting into new co-author disputes, new corrections, new clines replacing old arrows. The river, and the poets who sang to it while it still ran, were not waiting on the outcome.