The Mahabharata is a literary work, not a chronicle, and no excavation has confirmed the war it describes. Most historians accept that a memory of a real conflict among clans of the upper Ganga plain may lie under the poem, and that everything above that memory grew over roughly eight centuries of composition.

Is the Mahabharata a historical record?

No. The Mahabharata calls itself itihasa, a word usually rendered "history" but closer to "thus it was" - an account of the past told to make a moral argument, not to keep a register of events. It contains talking animals, weapons that answer to mantras, gods who father children, and a battle whose casualty figures run to millions. None of that is offered as reportage, and reading it as reportage misses what the text is doing.

The interesting question is narrower: is there a historical kernel? Did some conflict among ruling clans in the Ganga-Yamuna region leave a memory that bards later grew into a hundred thousand verses? That question is open, and the honest answer is that the evidence is suggestive and thin in equal measure.

Who wrote the Mahabharata?

Why does the text itself make dating hard?

The Mahabharata was not written once. It grew. The tradition preserves the memory of that growth: an original core called Jaya, an expanded Bharata, and finally the Mahabharata of a hundred thousand verses. Scholars reading the language, the metre, the legal ideas and the religious assumptions of different sections find the same layering, and generally place composition between roughly 400 BCE and 400 CE.

That span matters. A passage describing a city, a ritual or a weapon may reflect the world of the person who added it, not the world of the story. The Bhagavad Gita, the didactic mass of the Shanti and Anushasana parvas, and much of the Krishna theology are widely treated as later strata. So when the epic mentions an iron weapon or a grand walled capital, that detail cannot be used to date the events it narrates.

The eighteen parvas

What has archaeology found at Hastinapur?

The most cited excavation is B. B. Lal's work at Hastinapur in Uttar Pradesh in the early 1950s. Lal found a settlement with several occupation periods, the second of which produced Painted Grey Ware, a fine grey pottery with painted designs associated with the upper Ganga and Ghaggar valleys in roughly the twelfth to sixth centuries BCE. He also found evidence that a phase of the settlement ended in flood damage, which he connected with the tradition that Hastinapura was washed away by the Ganga and the Kuru capital moved to Kausambi.

The same Painted Grey Ware horizon turns up at other places the epic names, including sites identified with Indraprastha, Kurukshetra, Panipat, Tilpat and Baghpat. That is a real pattern and it is worth taking seriously: the epic's core geography matches a genuine archaeological culture in the right part of India.

What it does not show is a war. No mass grave, destruction layer, weapon cache or inscription at any of these sites has been tied to a battle between two coalitions. The pottery tells us people lived there; it does not tell us what they did.

What about Dwarka under the sea?

Krishna's city of Dwaraka is described in the epic as built on the coast of Saurashtra and finally swallowed by the sea. Marine archaeology off modern Dwarka and Bet Dwarka in Gujarat, led from the 1980s by S. R. Rao for the National Institute of Oceanography, recovered submerged stone blocks, what were reported as bastions and jetties, stone anchors and pottery.

The finds are real. Their interpretation is contested. Dates offered for the submerged structures have ranged from the second millennium BCE to the early centuries CE, the stone anchors are of types used over a long period, and several archaeologists have argued that some of the "walls" are natural formations or much later harbour works. Treat the identification of these remains with the epic's Dwaraka as a hypothesis, not a result.

Does any old inscription mention the war?

Yes, one famous one. The Aihole inscription in Karnataka, composed by the poet Ravikirti to praise the Chalukya king Pulakeshin II, dates itself twice: 3,735 years after the Bharata war, and in the 556th year of the Shaka era. The Shaka figure puts the inscription at about 634 to 635 CE, and running the other number backwards from it gives roughly 3102 BCE for the war.

This is important evidence, but of a specific kind. It shows that by the seventh century CE educated Indians treated the Bharata war as a fixed point in the past and used it as an era, in the same way a European charter might date itself from the founding of Rome. It is a record of belief about chronology, not an independent record of the event, and it stands some three and a half millennia after the date it assigns.

When do people say the war happened?

The traditional date is 3102 BCE, the year the Kali Yuga is held to have begun. It became standard through the astronomical tradition, notably Aryabhata in the late fifth century CE, and is the figure behind the Aihole inscription and behind most temple and pilgrimage literature since.

A long series of attempts has tried to recover a date from the epic's own sky. The text mentions eclipses, comets, the positions of planets, and Bhishma waiting for the winter solstice before dying. Researchers have fed these into planetarium software and produced dates including 3067 BCE, 2559 BCE, 1793 BCE, around 1478 BCE and around 1400 BCE. They disagree because the astronomical references are internally inconsistent, some are formulaic omens rather than observations, several belong to demonstrably late layers of the text, and the calculation depends on which references the investigator chooses to trust.

Most academic historians decline to assign a date to the war at all, and if they place anything, they place the society the poem remembers in the later Vedic period, broadly the twelfth to ninth centuries BCE, on the strength of the Kuru clan's prominence in Vedic literature and the Painted Grey Ware evidence.

The eighteen days of the war

What do historians actually agree on?

Four things, roughly. First, the Kurus were a real clan: they appear in the Rigveda and later Vedic texts as a dominant power of the upper Ganga-Yamuna region, and the Kuru-Panchala alliance is a historical political formation, not a novelist's invention. Second, the epic's geography is real and consistently mapped onto places that existed. Third, the text as we have it is a compilation spanning centuries and cannot be read as a single author's testimony. Fourth, nothing found in the ground so far confirms or refutes a great battle at Kurukshetra.

That leaves the sensible position: a genuine dynastic conflict may well sit at the bottom of the Mahabharata, the way a real siege may sit at the bottom of the Iliad, and the poem that grew over it is a work of imagination, argument and theology rather than a record of it. The epic is valuable because of what it thinks, not because of what it might once have reported.

Key facts

Archaeological culture: Painted Grey Ware, c. 1200-600 BCE
Key excavation: Hastinapur, B. B. Lal, 1950-52
Traditional date: 3102 BCE, start of the Kali Yuga
Oldest dated mention: Aihole inscription, c. 634 CE
Composition span: c. 400 BCE to 400 CE
Confirmed by excavation: Nothing about the war itself

Sources

B. B. Lal, "Excavation at Hastinapura and Other Explorations in the Upper Ganga and Sutlej Basins", Ancient India 10-11 (1954-55).
Romila Thapar, The Past Before Us: Historical Traditions of Early North India (2013).
Alf Hiltebeitel, Rethinking the Mahabharata (2001).
S. R. Rao, The Lost City of Dvaraka (1999), and the critical responses to it in Indian archaeological journals.
The Aihole inscription of Pulakeshin II, edited in Epigraphia Indica vol. VI.
https://en.wikipedia.org/wiki/Painted_Grey_Ware_culture

Where the guides summarise a disputed point, they say so. How this site is put together.