At its height, more than five million people lived in its cities — more than in ancient Egypt and Mesopotamia combined. We have spent a century excavating its ruins and we still cannot read a single word it wrote.
In 1922, two archaeologists working independently in what is now Pakistan made discoveries that would rewrite the history of human civilisation. John Marshall's team uncovered the buried city of Mohenjo-daro on the Indus River, while R.D. Banerji found similar ruins at Harappa, 600 kilometres to the north-east. What they had stumbled upon was the largest urban civilisation of the ancient world — one that had been lying undiscovered in plain sight beneath the soil of the subcontinent for three and a half millennia.
The Indus Valley Civilisation — also called the Harappan civilisation — flourished from roughly 3300 BCE to 1300 BCE, with its urban peak between 2600 and 1900 BCE. At that peak, it covered approximately 1.25 million square kilometres across what is now Pakistan, northwest India, and Afghanistan — an area larger than ancient Egypt and Mesopotamia combined. Its population at the height of its urban phase may have exceeded five million people.
And yet, despite more than a century of excavation, despite modern isotope analysis and satellite archaeology and machine-learning decipherment attempts, this civilisation remains stubbornly, fascinatingly mysterious. We have never deciphered its script. We do not know what language its people spoke. We do not know why its cities were abandoned. We do not even know what it called itself.
The physical remains of Harappan cities are remarkable by any standard. Mohenjo-daro, the best-excavated site, reveals a city planned with a sophistication that would not be matched in Europe for two thousand years. Streets run in a grid pattern oriented to the cardinal directions. Houses are built from standardised fired mud-brick in a consistent ratio of 1:2:4. Almost every house, including modest single-room dwellings, has access to a sophisticated drainage system — covered brick sewers running beneath the streets, connected to individual bathing platforms and latrines.
This is genuinely extraordinary. The Romans are often cited for their aqueducts and sewers, but Roman urban sanitation was largely public — baths and latrines shared by the community. The Harappans appear to have provided private sanitation infrastructure to individual households on a scale unprecedented in the ancient world.
The city also contains a massive Great Bath — a watertight brick structure 12 metres long, 7 metres wide, and 2.4 metres deep, with staircases at either end and what appears to have been changing rooms along the sides. Its function remains debated, but its careful construction — the brick floor sealed with natural tar and further lined with fired brick, all set in gypsum mortar — suggests it was used for ritual bathing rather than ordinary hygiene.
At Harappa and other sites, archaeologists have found standardised weights and measures consistent across the entire civilisation. A system of bronze weights used a binary-and-decimal hybrid — 0.05, 0.1, 0.2, 0.5, 1, 2, 5, 10, 20, 50, 100, 200, 500 — that appears to have governed trade across a geographic spread of over a thousand kilometres. This implies a level of administrative coordination and commercial infrastructure that is astonishing for the third millennium BCE.
"What strikes me most about the Harappans is not what we find in their cities — the drains, the standardised bricks, the sophisticated weights — but what we do not find. No grand palaces. No warrior kings' tombs. No evidence of militaristic conquest. They seem to have organised one of the largest urban societies in history without the apparatus of domination that we associate with ancient states." — Dr. Shantha Krishnaswamy, archaeologist, Deccan College Post-Graduate and Research Institute, Pune
The Indus script is one of the most tantalising unsolved puzzles in linguistics. Approximately 4,200 inscribed objects have been found — mostly small seals made of steatite, used for stamping goods or marking ownership. Each seal typically shows a brief inscription of between 5 and 26 characters, often accompanied by an animal figure: a unicorn-like creature with a single horn, a zebu bull, a tiger, a rhinoceros, an elephant.
The script has been studied by linguists, cryptographers, and computer scientists for a century. Machine-learning analyses using statistical methods borrowed from natural language processing have confirmed that the script shows the statistical signatures of a real language — it is not merely decorative, and it has internal patterns consistent with grammatical structure. But none of the decipherment attempts have produced a widely accepted result.
The fundamental problem is the absence of a bilingual inscription — something like the Rosetta Stone, which allowed Egyptian hieroglyphics to be decoded because the same text was written in hieroglyphics, Demotic, and Greek. Without a known-language parallel, deciphering an unknown script for an unknown language using inscriptions that average fewer than five symbols is extraordinarily difficult.
A contentious but interesting 2004 analysis by Rajesh Rao and colleagues, using computational methods, suggested the script shows the conditional entropy characteristics of a linguistic system rather than a non-linguistic system like DNA sequences or music notation. This supports the interpretation that the Indus script encodes a real language — but tells us nothing about what that language was.
Around 1900 BCE, something began to go wrong. The great cities of Mohenjo-daro and Harappa began to decline — populations shrank, civic infrastructure deteriorated, the standardised quality of construction fell, and the sophisticated drainage systems were no longer maintained. By 1300 BCE, the urban civilisation had essentially collapsed, with populations dispersing into smaller rural settlements.
The cause of this collapse has been debated for decades, and the honest answer is that we do not know. Three major hypotheses have been advanced, and all three are probably partially correct.
The most dramatic is the Aryan Invasion hypothesis — the idea that horse-riding warrior peoples from the Central Asian steppes swept through the northwest passes and destroyed the Harappan cities by force. This hypothesis was popular in the nineteenth and early twentieth centuries and has ideological dimensions that have complicated the scholarship. Physical anthropological evidence does not support mass violent destruction at most sites, and the genetic picture is complex. What is clear from recent ancient DNA studies published in Nature and Science is that there was significant steppe ancestry influx into South Asia around this period — but the timing and mechanism are contested.
The second hypothesis focuses on climate change. Paleoclimatological evidence — reconstructed from lake sediments, pollen records, and isotope analyses in cave formations — suggests that the monsoon weakened significantly in the region between 2200 and 1700 BCE, a period of global climate disruption that also coincided with the collapse of the Akkadian Empire in Mesopotamia and the Old Kingdom of Egypt. A weakened monsoon would have reduced agricultural yields and strained the food systems that supported dense urban populations.
The third hypothesis emphasises river change. Recent geological and remote-sensing work has identified a now-dry river system — likely the ancient Ghaggar-Hakra, which some scholars identify with the Vedic Sarasvati river — that once ran through the heart of the Harappan cultural zone. Evidence suggests this river dried up as tectonic shifts and monsoon changes altered the drainage pattern of the subcontinent, potentially severing agricultural water supplies from thousands of settlements simultaneously.
The most recent advances in understanding the Indus Valley Civilisation have come not from archaeology but from ancient genomics. Studies published in Cell and Science between 2018 and 2024 have analysed ancient DNA from sites across South Asia, shedding new light on the population movements of the period.
The emerging picture is complex but genuinely illuminating. The genetic ancestry of South Asians today includes at least three major components: an Ancestral South Indian (ASI) component related to the original hunter-gatherers of the subcontinent; an Ancestral North Indian (ANI) component that itself combines ancient Iranian-related ancestry with significant steppe ancestry from Central Asian pastoralists; and components related to early Southeast Asian populations.
The ancient Iranian-related ancestry — present in the region well before the steppe influx — is likely related to the farmers who built the Indus Valley Civilisation. This component is present throughout South Asia and in the Zagros mountain farmers of early Iran, suggesting deep ancestral connections between early Indus Valley populations and the agriculturalists of the greater Near East. The steppe ancestry influx, arriving around 2000-1500 BCE, correlates with the introduction of Indo-European languages to South Asia and aligns roughly with the Harappan urban decline period.
The Indus Valley Civilisation holds a peculiar position in South Asian national identity. In both India and Pakistan, it is invoked as evidence of ancient civilisational achievement — pre-dating the Vedic cultures that would later shape the region's religious and linguistic landscape. Political nationalism has sometimes distorted the scholarship, with some Indian researchers pushing interpretations that minimise the steppe ancestry component and others overclaiming indigenous continuity.
The science, stripped of its political freight, is remarkable enough on its own terms. A civilisation of five million people, operating across a geographic range larger than any other Bronze Age state, with urban planning that would not be surpassed in sanitation infrastructure for two thousand years — and we cannot read a single word it wrote. Its cities were abandoned peacefully enough that most of the decline happened without the dramatic destruction layers archaeologists find at sites sacked by armies. And somewhere in the undeciphered seals, in their unicorn-figures and careful pictographs, its own account of itself sits waiting.
The Indus Valley Civilisation is the ancient world's greatest unanswered question. And the answer, when it finally comes, will likely require advances in linguistics, climatology, genetics, and archaeology working together — the same interdisciplinary collaboration that the civilisation itself, with its standardised weights and long-distance trade networks, already seemed to understand.