Harappa: The Lost Cities of the Indus, the Script No One Can Read and the Fight Over Who We Are

Punjab, the 1850s. The British have just taken the land of the five rivers, and they are in a hurry to tie Lahore to Multan with an iron road. A railway needs ballast: hard broken stone or brick to bed the sleepers in. On the plain near the Ravi, the contractors find a gift. Mounds of old baked bricks, perfectly made, lying there for the taking.

So they take them. By one estimate, about 160 km of the line between Multan and Lahore ends up resting on bricks from a place the villagers call Harappa.

Nobody asks how old the bricks are. Nobody knows that the trains are rolling over a city some 4,500 years old, one of the largest of its age anywhere on earth.

Others had noticed. Charles Masson, a deserter from the East India Company’s army, wandered past in 1829 and described the ruins. In 1875 Alexander Cunningham, the first head of the Archaeological Survey of India, published a small stone seal from Harappa carved with a bull and a line of strange signs. He could not place it. Nobody could, for half a century.

This is the story of how that seal turned out to be a door into a lost world: a civilisation of planned cities, covered drains and ships bound for Iraq, with a script that still defeats every expert alive. It is also the story of a 100-year fight over whose ancestors they were.

In 1921, Daya Ram Sahni, a Punjabi archaeologist of the Archaeological Survey, began digging at Harappa. He found seals like Cunningham’s in layers far below anything anyone could date.

About 600 km to the south-west, in Sindh, another Indian archaeologist was puzzling over a mound on the Indus. Rakhal Das Banerji had visited Mohenjo-daro in 1919–20 because of a Buddhist stupa on top, a building of the early centuries CE. Then he found a flint scraper beneath it, and began to suspect that the stupa was the youngest thing on the hill. In 1922 he started digging.

Two men, two mounds 600 km apart, and the same strange seals coming out of both.

In June 1924, according to the archaeologist Supriya Varma, Sahni and Banerji sat down with their chief, John Marshall, the British Director-General of the Archaeological Survey, and laid the finds side by side. They matched. These were not two odd sites but one civilisation, spread across the whole Indus plain.

On 20 September 1924 Marshall announced it to the world in the Illustrated London News. Within a week the Oxford scholar Archibald Sayce wrote that the seals looked like ones found in Mesopotamia and Iran. That clue dated it. India’s known history, which had begun with the Mauryas, suddenly reached back two thousand years further.

Marshall’s name went on the announcement. Sahni, born in Bhera in Punjab, became in 1931 the first Indian Director-General of the Archaeological Survey, and the civilisation is called Harappan after the site he dug.

Walk through Mohenjo-daro today and the first thing you notice is the straight lines. The main streets run north–south and east–west, as if someone laid them out with string.

Then you notice the drains. Covered brick channels run along the streets, fed by pipes from houses, many of which had a small room for bathing. Archaeologists have counted more than 700 wells in the city, roughly one for every three houses.

On a raised mound stands the most famous structure of all: the Great Bath, about 12 m long, 7 m wide and 2.4 m deep, its bricks sealed with bitumen so it held water. Steps lead down at each end. It may have been used for ritual bathing; nobody knows.

What is missing is just as striking: no great palace, no temple that is clearly a temple, no statues of kings boasting of conquests. Who ran these cities is still a puzzle.

Mohenjo-daro means ‘mound of the dead’ in Sindhi; it is a later local name, and nobody knows what its own people called it. At its peak it may have held about 40,000 people, though scholars call that estimate weak. It became South Asia’s first UNESCO World Heritage Site in 1980.

Two small objects from the city became world-famous: a bronze girl just 10.5 cm tall, hand on hip, arm full of bangles, the ‘Dancing Girl’; and a bearded man carved in soapstone, the ‘Priest-King’, though no evidence shows priests or kings ruled here.

In 1947 Harappa and Mohenjo-daro went to Pakistan and their treasures were divided: the Priest-King stayed in Karachi, the Dancing Girl went to New Delhi. (Our long read on the Partition of Punjab tells the rest of that year.)

Harappa sits about 24 km west of Sahiwal, in Pakistan’s Punjab. When it was alive, the Ravi ran past its walls. Today the river has wandered about 8 km to the north, and wheat fields cover what was once the riverbank.

The digs there show how slowly a city grows. Farmers were living on the spot by about 3300 BCE, in what archaeologists call the Ravi phase. By the mature Harappan age, about 2600 to 1900 BCE, Harappa covered some 150 hectares and may have housed up to 23,500 people; other estimates go higher.

Some of the oldest marks anywhere in the subcontinent come from here: little clay and stone tablets of about 3300–3200 BCE scratched with signs like tridents and plants. Richard Meadow of Harvard, who directed the Harappa project, put it carefully: it is a big question whether these are ‘true writing’.

Harappa also shows the ending. After about 1900 BCE, a new way of burying the dead appears there, in a burial ground archaeologists named Cemetery H. The Cemetery H culture spread across Punjab, Haryana and western Uttar Pradesh, the late chapter of the Harappan world. Hold on to it: it comes back in the Aryan argument.

Around 2300 BCE, in what is now Iraq, the conqueror Sargon of Akkad boasted in an inscription that ships from three far-off lands moored at his capital: Dilmun, Magan and Meluhha. Dilmun is Bahrain. Magan is Oman. Most scholars believe Meluhha was the Indus civilisation.

Mesopotamian texts describe what came from Meluhha: carnelian, the blood-red stone the Indus bead-makers were famous for, gold, ivory, and fine timber carried in large ships. One Akkadian cylinder seal, now in the Louvre in Paris, belonged to a man called Shu-ilishu. Its inscription calls him an interpreter of the Meluhhan language. Somebody in Iraq was earning a living translating for traders from the Indus, more than 4,000 years ago.

Later tablets mention a ‘village of Meluhha’ in southern Iraq, perhaps a colony of Indus merchants. About twenty Indus seals have been found in Mesopotamia, at Ur, Kish and elsewhere.

At Lothal in Gujarat, builders made a large brick basin that many archaeologists see as one of the world’s earliest docks; others think it was a reservoir. After about 1760 BCE, the name Meluhha disappears from Mesopotamian records.

Thousands of these seals and tags survive, around 4,000 to 5,000 inscribed objects in all. On them are some 400 to 700 different signs, depending on whose count you trust.

The trouble is length: the average inscription is about five signs. And there is no Rosetta Stone, no text in both Indus signs and a known script. Most experts agree it was usually read from right to left. Beyond that, almost nothing is agreed.

More than 100 decipherments have been published since the 1920s, and they contradict each other. The Finnish scholar Asko Parpola and Mahadevan argued for a Dravidian language, an ancestor of the family that includes Tamil. The Indian archaeologist S. R. Rao argued for an Indo-Aryan language, related to Sanskrit.

In 2004 three Western scholars, Steve Farmer, Richard Sproat and Michael Witzel, went further. They argued the signs were not writing at all, but symbols that did not record a language. A 2009 study in Science by Rajesh Rao and colleagues answered that the patterns of the signs look like language. The fight goes on.

At Dholavira in Gujarat, archaeologists found ten large signs that had once been set on a wooden board near the city gate, each about 37 cm high. Nobody can read what this ‘signboard’ says.

In January 2025, Tamil Nadu Chief Minister M. K. Stalin announced a prize of one million US dollars for anyone who deciphers the script. As we write in October 2026, it has not been claimed.

Dholavira stands on Khadir Bet, an island in the salt desert of the Rann of Kutch. A local man, Shambhudan Gadhvi, kept asking officials to look at the ruins; J. P. Joshi of the Archaeological Survey officially recorded the site in 1967–68. From 1990 to 2005, R. S. Bisht led thirteen seasons of digging.

What they found was a city built around a single problem: water. Rain is rare in Kutch. Dholavira’s people dammed the two seasonal streams on either side of the city and cut a chain of stone reservoirs into the rock, some of them about 7 m deep. In 2014 archaeologists found a stepwell 73 m long.

Dholavira also left graves with no bodies in them. Of the funeral structures excavated, all but one were empty, perhaps memorials to people who died far away. On 27 July 2021 UNESCO named ‘Dholavira: a Harappan City’ a World Heritage Site.

About 150 km north-west of Delhi, in Haryana, the village of Rakhigarhi sits on top of one of the largest Harappan settlements ever found. It is often given as 350 hectares across eleven mounds, which would make it bigger than Mohenjo-daro; before 2014 most scholars put it at 80 to 100 hectares, and estimates still vary.

The Archaeological Survey began work there in 1963. Suraj Bhan of Kurukshetra University recorded it in 1969, Amarendra Nath of the Survey dug in 1997–2000, and Vasant Shinde’s team from Deccan College, Pune, dug from 2011 to 2016. They found paved streets, drains, a granary, bead workshops and a cemetery. In one grave a man and a young woman lie side by side; diggers nicknamed them the ‘love birds’.

By 2020 only about 5% of Rakhigarhi had been dug. Villagers have built about 152 houses on two of the mounds, and the Survey controls only a small part of the site. In February 2020, Finance Minister Nirmala Sitharaman announced that it would be developed as an ‘iconic site’, and digging resumed in 2021.

Its most famous find, though, was one skeleton. We will come to her.

In August 1947 the new border handed Harappa and Mohenjo-daro to Pakistan. India had lost the two sites that gave the civilisation its name, and its archaeologists went looking for Harappans on their own side.

They found them on the Sutlej. At Ropar, now Rupnagar, a young archaeologist called Y. D. Sharma began digging a mound in 1953, the first excavation of a Harappan site in independent India. He found a sequence of six cultures stacked one on another, from Harappan pots at the bottom to medieval ones at the top, and a Harappan cemetery. One grave held a dog buried with its master.

Nearby, at Bara, about 6 km south-west of Ropar, Sharma found a distinctive brownish-red pottery, painted and incised. It gave its name to the Bara culture, a regional and largely late phase of the Harappan world.

Further south, at Sanghol in Fatehgarh Sahib district, digs, above all in 1986, uncovered a mound used from the Harappan Bara phase to the Gupta age. It is best known for a Buddhist stupa of the Kushan era, but its lowest layers yielded lemon seeds, the only evidence of citrus that far west at the time.

Here is the uncomfortable part. Seventy years later, the full excavation reports for Ropar and Sanghol have still not been published, a delay India’s auditor, the CAG, flagged in 2013 and again in 2022. Ropar’s museum had no curator when The Tribune visited in 2024. The archaeologist Disha Ahluwalia wrote the same year that Punjab’s archaeology is ‘stuck in the 1950s’ and that small sites around Chandigarh are being levelled.

No army burned Mohenjo-daro. No volcano buried Harappa. The great Indus cities died slowly, like a town whose young people leave one by one.

From about 1900 BCE the signs of decline spread. By about 1700 BCE most of the big cities were empty. Yet the people did not vanish. In what is now India, the archaeologist Gregory Possehl counted 218 known sites before 1900 BCE and 853 after. People moved east and into smaller villages, towards Haryana, the Ganga–Yamuna plain and the foothills.

Why? The leading explanation is climate. Around 4,200 years ago, a mega-drought hit much of the world, from Egypt to Mesopotamia; scientists call it the 4.2-kiloyear event. A 2012 study in the journal PNAS led by Liviu Giosan of the Woods Hole Oceanographic Institution argued that the Harappans had relied on monsoon-fed rivers and floods, and that as the monsoon weakened those rivers became seasonal or dried up.

There was another explanation, once very popular. It was wrong.

In the 1940s Mortimer Wheeler, digging at Harappa and Mohenjo-daro, pointed to skeletons lying in Mohenjo-daro’s streets and to Rigveda hymns about Indra, the war god who smashed forts. He blamed the Aryans for a massacre. ‘Indra stands accused,’ he wrote.

In 1964 the American archaeologist George Dales took the case apart: the 37 skeletons came from different periods, and there was no burnt destruction layer and no last stand on the citadel. In 1994 Kenneth Kennedy showed the marks on the skulls came from erosion, not weapons. Wheeler’s massacre is a myth.

The Rigveda praises a mighty river, Sarasvati, flowing from the mountains to the sea. By the time of later texts it disappears into the desert. Many Harappan sites line a dry riverbed called the Ghaggar in Haryana and Punjab and the Hakra across the border. Was that the Sarasvati?

It matters because if the Ghaggar-Hakra was the Sarasvati, and the Vedic hymns knew it as a great river, then some argue the hymns are as old as the Harappan cities, and the Vedic people might be the Harappans themselves. That is why some writers prefer the name ‘Sindhu-Sarasvati civilisation’.

The 2017 study found something both sides can use. The Harappans did build along a former river: a valley the Sutlej had abandoned, which still caught monsoon water and was safer from floods. The river was real. Whether it was the Sarasvati of the hymns, and whether it was still a great river when the hymns were sung, is the argument.

Few questions in Indian history start more fights: where did the Rigveda’s composers come from?

The Rigveda is in an early form of Sanskrit, an Indo-European language. Its cousins include Persian, Greek, Latin and English. Its hymns name Punjab’s rivers: Shutudri for the Sutlej, Vipash for the Beas, Parushni for the Ravi. Most scholars date it to roughly 1500–1200 BCE, after the Indus cities had faded.

The colonial version spoke of an ‘Aryan invasion’, conquerors crushing the natives. Almost no serious scholar holds it today. The debate now is between two very different pictures.

For decades neither side could prove its case. Language leaves no bones. Then geneticists learned to read DNA from bones 4,000 years old.

She was buried in the cemetery at Rakhigarhi sometime between about 2800 and 2300 BCE. Scientists know her only as I6113. Of 61 skeletons screened, hers was the only one with usable DNA.

In September 2019, two papers appeared together. The one in the journal Cell, led by Shinde with the Harvard geneticist David Reich and others, described her genome. The other, in Science, led by Vagheesh Narasimhan, compared hundreds of ancient people from South and Central Asia.

What the Rakhigarhi woman showed is not in dispute. Her ancestry came mostly from a population related to ancient Iranian hunter-gatherers, mixed with ancestry related to South Asia’s oldest hunter-gatherers. She had no detectable ancestry from the Steppe herders of Central Asia. Reich said ancestry like hers is ‘the primary ancestry source in South Asia today’. Most South Asians, in other words, descend in large part from people like her.

What it means is very much in dispute.

Our study proves that there has been no foreign induction in our civilisation.Vasant Shinde, lead author of the Cell paper, to The Week, September 2019

Shinde told reporters the Aryan invasion theory was ‘based on very flimsy ground’, that the Vedic culture and the Harappan were the same, and that ‘We are descendants of the Harappans.’ He also said some Steppe migration did happen, but the newcomers mixed in without conquering. His co-author Niraj Rai said an Out of India direction ‘could be the future direction of our research’.

Others read the same papers the other way. The Science paper found that Steppe ancestry did reach South Asia, in the first half of the second millennium BCE, after the woman of Rakhigarhi lived. Reich has written that Steppe ancestry is more pronounced in sections of traditional upper castes. The Cell paper itself called a route for Indo-European languages from Eastern Europe through Central Asia in that period plausible. Tony Joseph said there was ‘not even a suggestion’ in the genetic papers of large-scale migration out of India.

The argument moved from journals into classrooms. In 2024 NCERT revised the Harappan chapter in its Class 12 history textbook, Themes in Indian History Part I. According to reports, the new text cites the Rakhigarhi research as showing an unbroken genetic line over 5,000 years and argues that this rules out large-scale immigration of the so-called Aryans. Critics, among them The Wire, said it ignored the later Steppe ancestry in the Science paper. Reports said the new Class 6 textbook uses the name Sindhu-Sarasvati civilisation.

A civilisation nobody can read is a magnet for stories. Here are the most common ones, set against the record.

Strip away the shouting and the record is more remarkable than the myths. On the plains of Punjab, Sindh, Haryana and Gujarat, people built cities with drains before most of the world had cities at all. They traded with Iraq and wrote in signs we still cannot read. When the rains failed they did not fight to the last; they moved, and their descendants farmed on.

The Aryan argument will not be settled by slogans. So far the Rakhigarhi woman tells us the Harappans were not Steppe people and that most South Asians carry their blood. The companion study tells us people with Steppe ancestry arrived later and mixed in. Both can be true at once, and the honest reader should hear both.

And one thing is plain. The first Indian to dig Harappa was a Punjabi, and the first site independent India dug was on the Sutlej. Seventy years on, the reports from Ropar and Sanghol are still unpublished. Punjab’s oldest story deserves better.