Where We Left Off
A short recap for anyone starting here, and a reminder of what the coloured boxes mean.
Part 1 built the stage. Two continents crashed slowly into each other and threw up the Himalayas. Rivers ground those mountains into powder and dumped the powder on the plains, flood after flood, until the soil was kilometres deep. The result was flat, endlessly fertile land with water all year round, sitting directly behind the only easy way into India.
Then it followed the first people. Stone tools in the Soan valley of the Potohar Plateau, made by a human species that was not yet us. A long, thin, badly-recorded stretch of hundreds of thousands of years. And finally, around nine thousand years ago, farming — mud-brick houses at Mehrgarh, wheat and barley, herded goats, and grain kept in jars.
Grain in a jar is where civilisation starts. Store food and you can feed people who are not farmers. Feed people who are not farmers and you get potters, builders, priests, soldiers, accountants. That is the recipe, and Part 1 ended with small farming villages spreading out across Punjab, doing exactly that.
This part is what those villages turned into.
What the boxes mean
Same system as Part 1. If you have read it, skip ahead.
Blue explains a hard word, right where the word appears.
Example — Archaeology: the study of the human past through the physical things people left behind. Buildings, pots, bones, rubbish. An archaeologist digs and interprets. They are not treasure hunters; most of what they find is broken and worthless, and that is the useful part.
Green turns a number or an idea that is too big or too strange to picture into something ordinary.
Brown shows the evidence itself — the brick, the bone, the seal, the chemical trace — and is honest about what that evidence can and cannot prove.
Orange means serious people disagree. Both sides get their best case made for them, then a plain statement of where things actually stand and what evidence would settle it.
Purple digs out something everybody has quietly taken for granted without proving. Part 2 has more of these than Part 1, because almost everything about Harappa is interpretation resting on very thin evidence.
Dark boxes close each chapter with the one thing you must take away, in the plainest words I can find.
Read only the dark boxes and you will still have the whole story.
A warning specific to this part
Part 1 was mostly geology, and geology is comparatively calm. Rocks do not have descendants who mind what you say about them.
Part 2 is different. Harappa sits at the centre of live political arguments in India today — about who was here first, about whether Indian civilisation is home-grown or partly imported, about what to call things. Some of the loudest claims you will meet online about this civilisation are made by people who need a particular answer to be true.
I will show you where the evidence is strong, where it is thin, and where honest researchers disagree for honest reasons. Where a claim is popular but weakly supported, I will say so — in whichever direction that cuts.
1How a Whole Civilisation Was Lost and Found
For three thousand years nobody knew it had existed. No Indian text remembered it. No Greek writer mentioned it. Then a railway line was built, and the workers needed something to put under the tracks.
1.1 — Bricks for a railway
In the 1850s the British were laying a railway line from Lahore to Multan, straight across Punjab.
A railway needs ballast — a bed of hard broken material under the sleepers to spread the weight and let water drain. On the flat plain of Punjab there is no stone. There is only soil, kilometres of it. Stone would have to be carried in from far away, at great cost.
Then somebody noticed the mounds.
Near a village called Harappa, on the old bed of the Ravi, stood large low hills of rubble. Local people had been quietly taking bricks from them for generations to build their own houses. The bricks were hard, well-fired, remarkably uniform, and there were millions of them.
So the engineers — the brothers John and William Brunton — used them. Around a hundred and sixty kilometres of railway track was laid on a bed of crushed ancient brick.
Think about what that means. To get enough ballast for a hundred and sixty kilometres of railway, you need an enormous quantity of brick.
Those bricks came out of a city. A city so large that it could be ground up and spread under a railway for a hundred and sixty kilometres, and there was still enough left in the ground for archaeologists to dig for the next hundred and fifty years.
Nobody at the time understood what was being destroyed. And here is the uncomfortable part: we will never know what was in those bricks. Seals, inscriptions, whole buildings. Under a railway.
1.2 — The people who suspected
A few people did notice something was odd.
In 1829 a British deserter and wanderer named Charles Masson passed through and wrote a description of the ruins at Harappa. He assumed they were connected to Alexander the Great, because that was the only ancient history of Punjab anyone in Europe knew.
In the 1850s and again in 1873, Alexander Cunningham, the first head of the Archaeological Survey of India, visited Harappa. He found a seal — a small square of stone with an animal on it and some marks he could not read. He published it. He assumed it was foreign, because the marks matched no Indian script he knew.
Both men looked directly at the evidence and interpreted it inside the framework they already had. Neither imagined that Indian civilisation might be thousands of years older than anyone thought. That possibility was simply not available to them.
This is worth stopping on, because it is a permanent danger and not just a nineteenth-century one.
Cunningham held a seal from a lost civilisation in his hand and concluded it must be foreign. Why? Because he had a mental map of Indian history that started with the Vedas, and this did not fit on that map. Rather than question the map, he questioned the object.
The assumption was: what I already believe about the past is roughly the right shape. New evidence gets fitted into that shape, and if it will not fit, it gets explained away as an oddity or an import.
Everyone does this. You do it. I do it. The only defence is to notice when you are working hard to explain a piece of evidence away, and ask whether the map might be wrong instead of the evidence.
1.3 — 1921 and 1922
The real discovery came in two seasons, and it was made by Indian archaeologists.
In 1921, Daya Ram Sahni began proper excavation at Harappa in Punjab.
In 1922, Rakhal Das Banerji was excavating a Buddhist site far to the south, in Sindh, at a place the locals called Mohenjo-daro — which means "mound of the dead". Digging below the Buddhist layers, he hit something much older. Brick buildings. Streets. Seals like Cunningham's.
The same seals. Six hundred kilometres apart.
That was the moment. One strange mound is a curiosity. Two strange mounds six hundred kilometres apart, with the same script, the same brick sizes and the same seal designs, is a civilisation.
In 1924 John Marshall, then head of the Archaeological Survey, announced it to the world in the Illustrated London News.
Imagine the shock. Before 1924, the standard account of Indian history began with the Vedas, at perhaps 1500 BCE, and everything before that was a blank.
Overnight, that blank filled with brick cities, plumbing, standardised weights and a writing system — a thousand years older than the Vedas, and geographically in the same place.
It is as if you had traced your family back three generations, and then found a locked room in your own house containing forty generations you had never heard of.
India went, in a single announcement, from having a long history to having one of the three oldest urban civilisations on Earth. And every argument in this part flows from the fact that we still cannot read a word those people wrote.
1.4 — What do we call it?
This sounds like a trivial question. It is not. The name you use tells people which side you are on, which is exactly why the argument exists.
Three names are in common use, and each carries a claim.
The oldest and most widely used name internationally. It names the civilisation after the river system where the first great sites were found. Objection: a very large number of sites — arguably the majority — are not in the Indus valley at all. They are along the dry Ghaggar-Hakra channel to the east, and along the coast of Gujarat. The name describes only part of the map.
Names it after the first site properly excavated, following the standard archaeological convention of naming a culture after its type site. Objection: Harappa is only the first site because of where the British happened to run a railway. It is not the largest, the oldest, or the most important. The name records an accident of colonial engineering.
Adds the name of the river described in the Rigveda, which many identify with the dry Ghaggar-Hakra channel where so many sites cluster. Argument for: it fixes the geographical inaccuracy of "Indus Valley" and acknowledges where most sites actually are. Objection: it imports a name from a religious text composed centuries later, and quietly assumes both that the Saraswati was this river and that the Harappans were connected to the people who wrote the Rigveda. Both are contested — see Chapter 8. Critics argue the name smuggles in a conclusion before the argument has been had.
What I do in this series: I use Harappan, because it is the most neutral of the three — it commits me to nothing except that we dug at Harappa first. That is a choice about caution, not a verdict on the Saraswati question, which I take up properly in Chapter 8.
The honest point: the naming argument cannot be settled by evidence, because it is not really about evidence. It is about which claim you want the name to carry. Recognising that is more useful than picking a winner.
A whole civilisation was completely forgotten for three thousand years. No text remembered it.
It was found because British engineers ground up its bricks for railway ballast, and somebody eventually asked where the bricks came from.
The real discovery was 1921–22, by Daya Ram Sahni at Harappa and R. D. Banerji at Mohenjo-daro. Two sites, six hundred kilometres apart, with identical seals — that is what proved it was a civilisation and not a curiosity.
Its name is still argued over, because the name you choose carries a claim about who these people were.
2The Size of the Thing
Most people picture two cities. It was more than a thousand settlements spread across an area larger than Egypt and Mesopotamia put together — and a large share of them were in Punjab.
2.1 — How big?
The Harappan world covered roughly 1.5 million square kilometres.
That is bigger than ancient Egypt and ancient Mesopotamia combined — the two civilisations everybody has heard of.
It is roughly the size of modern France, Germany, Italy, Spain and Britain added together.
It stretched from the coast of the Arabian Sea in the south to the foothills of the Himalayas in the north, and from the border of Iran in the west to near Delhi in the east.
And almost nobody outside South Asia has heard of it, because we cannot read what they wrote. Civilisations that left readable records get remembered. This one did not.
More than a thousand Harappan settlements have been identified, and the number keeps rising as surveys continue. The great majority are villages and small towns. Only a handful were large cities.
2.2 — The three phases
Archaeologists split it into three stretches. The dates are approximate and researchers argue over the edges.
| Phase | Roughly when | What it looks like |
|---|---|---|
| Early Harappan | c. 3300 – 2600 BCE | Farming villages and small walled towns. Regional pottery styles. Some planning, some craft specialisation, early trade. No large cities yet, no script. |
| Mature Harappan | c. 2600 – 1900 BCE | The famous period. Big planned cities, standard bricks, standard weights, the script, seals, long-distance trade, drainage. Roughly seven hundred years. |
| Late Harappan | c. 1900 – 1300 BCE | Cities decline and empty. Standards break down. The script disappears. People are still there — but living differently, in smaller settlements, spreading east. |
Notice the shape of that. It is not a sudden appearance and a sudden death. It is a long build-up, a seven-hundred-year peak, and a long unwinding. Chapter 9 is about that unwinding.
Seven hundred years at the peak. Put that next to things you know.
The Mughal Empire lasted about three hundred years. The British were in India for roughly two hundred. The United States has existed for around two hundred and fifty.
The mature Harappan period lasted longer than all three of those put together. Somebody born at the start of it and somebody born at the end were separated by more time than separates us from the Delhi Sultanate.
2.3 — The big cities
| Site | Where it is now | Why it matters |
|---|---|---|
| Mohenjo-daro | Sindh, Pakistan, on the Indus | The best-preserved city. The Great Bath, the grid streets, the "priest-king" statue and the dancing girl all come from here. |
| Harappa | Sahiwal district, Punjab, Pakistan, on the old Ravi | The type site. Heavily damaged by brick-robbing, but the longest-running excavation and the deepest sequence of layers. |
| Rakhigarhi | Hisar district, Haryana, India | Possibly the largest site of all — its full extent is disputed. Source of the only Harappan ancient DNA sequenced so far. |
| Dholavira | Kutch, Gujarat, India | Built of stone rather than brick. Spectacular water-storage system. The famous ten-sign "signboard". |
| Ganweriwala | Cholistan, Punjab, Pakistan | A very large site on the dry Hakra channel. Barely excavated — it sits in desert with almost no water supply for a dig. |
| Kalibangan | Rajasthan, India, on the Ghaggar | The ploughed field from before the city was built. Disputed "fire altars". |
| Lothal | Gujarat, India | The bead workshop and the great brick basin argued over as a dockyard. |
| Chanhudaro | Sindh, Pakistan | A craft town rather than a city. Bead-making on an industrial scale. |
2.4 — The Punjab sites
This is a history of Punjab, so it is worth being precise about which of this belongs to Punjab.
A great deal of it. Both Punjabs, on both sides of the modern border.
- Harappa itself is in Pakistani Punjab, in Sahiwal district, on the dried old course of the Ravi — one of the five rivers from Part 1.
- Ganweriwala is in Cholistan, in southern Pakistani Punjab.
- Ropar (Rupnagar) in Indian Punjab, on the Sutlej, was the first Harappan site excavated after independence, in 1953.
- Sanghol in Fatehgarh Sahib district, Indian Punjab, is an important Late Harappan settlement.
- Bara, near Ropar, gives its name to a whole Late Harappan regional style found across Punjab — "Bara ware".
- Banawali and Rakhigarhi are in Haryana, which was part of Punjab province until 1966.
- Kalibangan is in Rajasthan on the Ghaggar, immediately beyond Punjab's edge.
So when this part talks about Harappan cities, it is not describing somebody else's history that happened nearby. The type site of the entire civilisation stands on the bank of one of Punjab's five rivers.
How do you find a buried city? It is less mysterious than it sounds.
Mounds. When mud-brick buildings collapse and are rebuilt on top, over and over for centuries, the ground level rises. A long-occupied site becomes a low artificial hill — a tell or, in this region, a daro or theri. On a dead-flat plain, a hill is visible from a long way off and is not natural.
Surface scatter. Ploughing and rain bring broken pottery to the surface. An archaeologist walks a field, picks up sherds, and reads the style. Harappan pottery is distinctive.
Satellite and aerial imagery. Buried walls affect how well crops grow above them, so ancient street plans can show up as faint patterns in a field from the air. Old river channels show up the same way.
The limits, honestly: this method finds sites that are still visible. Anything flattened by centuries of ploughing, buried by river silt, or now under a modern town is invisible. Remember survivorship bias from Part 1 — the map of Harappan sites is a map of what has survived and been looked for, not a complete census.
The Harappan world covered about 1.5 million square kilometres — larger than ancient Egypt and Mesopotamia combined — with more than a thousand known settlements.
Three phases: Early (3300–2600 BCE), Mature (2600–1900 BCE, the famous cities), Late (1900–1300 BCE, the unwinding).
The mature period lasted seven hundred years — longer than the Mughals, the British Raj and the United States put together.
Harappa itself stands on the old bed of the Ravi, in Punjab. This is Punjab's own history, not a neighbour's.
3Walking Into a Harappan City
Straight streets on a compass grid. Bricks made to one ratio across a thousand kilometres. A covered sewer under the road and a bathroom in an ordinary house. Then imagine the year is 2500 BCE.
3.1 — The streets
Most ancient cities grew the way villages grow. A path here, a house there, another house squeezed beside it, a lane that bends around somebody's wall. Look at the old quarter of almost any city in the world and you will see it — a tangle that made sense to whoever built each piece.
Harappan cities do not look like that.
The main streets run straight, cross at right angles, and are aligned roughly north-south and east-west. Between them are blocks of housing, cut through by narrower lanes. Some main streets are around ten metres wide — wide enough for two carts to pass.
A city like that cannot grow by accident. Somebody decided the layout before the buildings went up.
Grid plan: a town laid out on straight lines crossing at right angles, like graph paper. Rare in the ancient world and normally a sign of central planning.
Cardinal alignment: lined up with north, south, east and west rather than pointing any old way. Doing this deliberately means somebody was measuring against the sun or the stars.
Most of the larger cities have two parts: a raised western section built on a brick platform, usually called the citadel, and a larger lower town to the east where most people lived.
Be careful with the word "citadel". It was chosen by early excavators who were thinking of forts and palaces. What is actually up there is public buildings — the Great Bath, big halls, large structures whose purpose is unknown. Not a palace. Not, as far as anyone can show, a barracks. The name has stuck and it quietly shapes how people picture the place.
3.2 — The bricks
Now the detail that says the most about how this society worked.
Harappan bricks are made to a fixed proportion: the length is four units, the width two, the thickness one. A ratio of 1:2:4.
This ratio holds at Harappa in Punjab. It holds at Mohenjo-daro in Sindh, six hundred kilometres away. It holds at Dholavira in Gujarat and at Kalibangan in Rajasthan.
Imagine every brick made in every town from Amritsar to Karachi, over seven centuries, coming out to the same proportions.
No telephone. No printing. No paper standards office. Just — everyone doing it the same way, for longer than the time between Guru Nanak and today.
And they were right to. A 1:2:4 brick can be laid in an interlocking pattern where the joints never line up between courses, which makes a much stronger wall. This is not decoration. It is engineering knowledge, and somebody was making sure it was followed.
The same is true of the weights, which we come to in Chapter 4. Something was enforcing standards over an enormous area for a very long time. What that something was is the great mystery of this civilisation.
3.3 — The drains
This is the part that startles people most.
An ordinary Harappan house — not a rich house, an ordinary one — commonly had:
- A bathing platform: a floor of tightly fitted brick, slightly sloped, often with a raised edge.
- A drain from that platform, running through the house wall.
- A connection to a covered drain running under the street outside.
- Often a private well in the courtyard, or shared access to one nearby.
The street drains were built of brick, covered with brick or stone slabs so they were not open sewers, and fitted at intervals with inspection openings — the ancient equivalent of a manhole — so they could be cleaned out. Some had settling pits, where solids collected and could be removed.
Mohenjo-daro alone is estimated to have had around seven hundred wells.
Here is the comparison that should stop you.
In 1858, London had a summer so foul it is remembered as the Great Stink. Human waste went into the Thames, and the Thames was the drinking water. Cholera killed tens of thousands. Parliament had to hang curtains soaked in chemicals over the windows because the smell coming off the river was unbearable. Only then was a proper sewer system built.
The Harappans had covered street sewers with inspection points, connected to bathrooms in ordinary houses, more than four thousand years earlier.
Not "advanced for their time". Better than the capital of the largest empire on Earth, in the age of steam.
Careful here, because the drains tempt everyone into a conclusion that does not follow.
The reasoning usually runs: good plumbing means an advanced society, and an advanced society means an equal, peaceful, well-governed one.
That is a leap. Plumbing is engineering. It tells you people understood water, gradient and brick. It does not tell you whether they were kind, whether women had rights, whether anybody was enslaved, or whether the people who dug the drains were treated well.
Rome had superb plumbing and a slave economy. Nineteenth-century Britain built magnificent sewers while running an empire by force.
The assumption is that technical sophistication and moral decency travel together. History gives no support for that whatsoever. The Harappans may well have been a decent society — the absence of weapons and fortifications is a real point in their favour, as Chapter 4 discusses. But that has to be argued from its own evidence, not read off the drains.
3.4 — The Great Bath
On the citadel mound at Mohenjo-daro sits the most famous single structure of the civilisation.
It is a rectangular sunken tank, roughly twelve metres long, seven metres wide and about two and a half metres deep, with brick steps down into it at each end. Around it runs a paved courtyard and a set of small rooms.
The construction is the impressive part. The floor and walls are made of finely cut brick laid in gypsum mortar, and behind the brick there is a thick layer of bitumen — natural tar — to waterproof it. There is a drain to empty it and a nearby well to fill it.
Bitumen: a thick black natural tar that seeps out of the ground in some places. It is waterproof and sticky. The Harappans used it as a sealant — the same basic idea as the tar on a modern flat roof.
What was it for? Nobody knows. It is usually assumed to be for ritual bathing, because that is what a large public tank next to important buildings suggests and because ritual bathing is important in later South Asian religion.
But notice what that reasoning does: it explains a Harappan building using a religious practice recorded thousands of years later. It may be right. It is not evidence. Chapter 7 comes back to this problem properly.
3.5 — The building that may not be a granary
At both Mohenjo-daro and Harappa there are large brick structures with a distinctive design: a solid platform crossed by narrow channels, with sockets that once held wooden beams. Early excavators — Marshall, and later Mortimer Wheeler — identified them as state granaries, with the channels providing air circulation to keep grain dry, and read them as evidence of a taxing, redistributing central authority.
The identification went into every textbook. It is now seriously doubted.
The design makes sense for grain storage: a raised, ventilated wooden floor keeps grain off damp ground, which is exactly what you need in a monsoon climate. The buildings are large, centrally placed and clearly public. Every comparable early civilisation had central storage — Mesopotamia and Egypt both did — and a city of tens of thousands must have had some way of holding a food reserve.
No grain has been found in them. Not carbonised grain, not grain impressions, not the storage jars you would expect, not the seed of the weeds that always come with stored grain. Archaeologists have looked hard. In a building used for centuries to store grain, some trace would normally survive — and at other sites in the region, actual grain storage does leave such traces. The identification came from what the excavators expected a great civilisation to have, not from what was in the building.
What would settle it: chemical residue analysis of the floors and soil, looking for the specific traces starch and plant matter leave behind. This is now technically possible and is worth doing.
Why this matters beyond the building: the granary reading was used as a main plank of the argument for a strong central state collecting taxes in grain. Remove the granary and one of the better arguments for Harappan central government goes with it. That is the subject of the next chapter.
3.6 — Dholavira and the water
One more site deserves its own moment, because it shows how far the planning went.
Dholavira sits on an island in the Rann of Kutch, in Gujarat. It is a desert. There are two seasonal streams and very little rain.
So the Harappans built a city that harvested and stored water on an extraordinary scale. They dammed the streams, and cut a series of enormous reservoirs into the rock — some more than seventy metres long and several metres deep — connected by channels, with steps down into them. A large part of the walled city area is given over to water storage.
They also built in stone rather than mud brick here, because stone was what the local geology offered.
That last point is important and often missed. These were not people mechanically following a template. The plan adapted to the place: brick on the alluvial plains where there is no stone, stone in Kutch where there is, massive water storage where water is scarce. Standardised where standards helped, local where local made sense.
Harappan cities were laid out on a grid, aligned to the compass, and planned before they were built.
Bricks were made to a fixed 1:2:4 ratio across a thousand kilometres and seven hundred years. Something was enforcing that.
Ordinary houses had bathing floors and drains connected to covered street sewers with inspection points — better sanitation than London had in 1858.
The famous "granary" was probably not a granary. No grain has ever been found in it.
Good plumbing proves good engineering. It does not prove a good society. That has to be argued separately.
4The Civilisation With No King
Egypt left pyramids full of dead pharaohs. Mesopotamia left king lists and boastful inscriptions. Harappa left a thousand cities and not one ruler anybody can name. This is the deepest puzzle in the whole subject.
4.1 — What is missing
Start with what every other early civilisation has, and Harappa does not.
| Normally you find | At Harappan sites |
|---|---|
| Palaces — a very large residence, far grander than any other building | None identified. Houses vary in size, but there is no building that is obviously a ruler's home. |
| Royal tombs — a burial loaded with gold, weapons, sacrificed servants | None. Burials are modest and fairly similar to each other. Some have more pots than others. Nothing like the tomb of a king. |
| Monumental temples | None certainly identified. There are large public buildings; none can be shown to be a temple. |
| Royal images — statues and carvings of rulers, usually oversized | Almost no large sculpture at all. The few human figures are small — the largest famous one is about 18 cm tall. |
| Inscriptions boasting of conquests | Nothing readable, and nothing that looks like a monumental public inscription — with the possible exception of the Dholavira signboard. |
| Heaps of weapons | Very few. Some spearheads and arrowheads, mostly light and often interpreted as hunting gear. No swords worth the name. No armour. No dedicated military architecture that everyone agrees on. |
| Battle scenes in art | Essentially none. No captives, no rows of prisoners, no victory scenes — all of which are standard subjects in Egyptian and Mesopotamian art. |
Compare it directly.
From Egypt we have the names, faces, mummified bodies and family trees of dozens of pharaohs, plus the pyramids they built to be remembered by.
From Mesopotamia we have king lists going back centuries, law codes with a named king's name on them, and carvings of rulers standing on piles of enemies.
From Harappa — a civilisation larger than both, lasting seven hundred years at its height, with a thousand settlements — we have not one name of one person.
Not a king. Not a priest. Not a merchant. Nobody.
4.2 — The statue we call a priest-king
There is one object everyone reaches for at this point. A small carved stone figure, about eighteen centimetres tall, found at Mohenjo-daro in 1927. A bearded man with a headband, an armband, and a patterned cloak over one shoulder. His eyes are narrowed, looking down.
It is on the cover of half the books about this civilisation. It is usually called the "Priest-King".
Now, honestly: there is no evidence he was a priest, and no evidence he was a king.
The name was given by early excavators who were thinking of Mesopotamia, where priest-kings were a real institution. They saw a dignified male figure with ornaments and reached for the nearest category they knew.
This is one of the most instructive mistakes in archaeology, because you can watch it happen.
An object is found. It has no label. Somebody names it using a category from a different civilisation. The name goes into the literature. Then, decades later, people cite the "Priest-King" as evidence that the Harappans had priest-kings.
The evidence and the interpretation have swapped places. A guess became a name, and then the name became a fact.
The same has happened to the "citadel" (which may not be a fortress), the "granary" (which probably was not one), the "Great Bath" (which may not have been for ritual), and the "mother goddess" figurines in Chapter 7.
When you read about Harappa, mentally strip the names off and ask what the object actually is. In this case: a small carved stone figure of a bearded man wearing ornaments. That is all we know.
4.3 — The weights
Against all that absence, here is something startlingly present.
Harappan sites produce large numbers of small cubes of stone — usually chert, beautifully finished. They are weights.
The system runs in doubling steps: 1, 2, 4, 8, 16, 32, 64 — and then shifts to larger multiples for heavier weights. The smallest are tiny, under a gram, suitable for weighing gold or gemstones. The largest are heavy enough to need a rope handle.
And they are accurate. Weights from sites hundreds of kilometres apart match each other closely.
This one is refreshingly direct: archaeologists collect the stone cubes and weigh them on modern precision scales.
When you plot the results, they do not scatter randomly. They cluster tightly around a repeating series of values, and the ratios between the clusters are the doubling series above.
Do the same for weights from Harappa in Punjab, Mohenjo-daro in Sindh, Lothal in Gujarat and Kalibangan in Rajasthan, and the clusters land in the same places.
This is about as hard as evidence gets in archaeology. It is a physical measurement, repeatable by anyone, from objects found independently at many sites. There is no interpretation involved.
Sit with what that implies. A merchant in Punjab and a merchant six hundred kilometres away in Sindh could trade by weight and both be confident of the measure. That requires the standard to be made, distributed, checked and trusted — for centuries.
You cannot have that by accident. Somebody was running something.
4.4 — So who ran it?
Here is the central problem stated plainly. The Harappans had enough organisation to enforce brick ratios and weight standards across a subcontinent for seven hundred years. And they left no trace of a ruler.
There is no agreed answer. Five serious proposals are on the table.
One authority ruled the whole area, which is why standards are uniform. For: the uniformity is very hard to explain otherwise. Against: a state that big normally advertises itself — palaces, royal art, propaganda. Harappa has none. It would be the only known empire in history that never bragged.
Mohenjo-daro, Harappa, Dholavira, Rakhigarhi and Ganweriwala each ran their own territory, sharing a common culture the way Greek city-states did. For: fits the lack of a single capital, and there are real regional differences in pottery and layout. Against: independent states usually fight each other, and there is little evidence of that. It also does not explain why the standards match so exactly.
Power lay with trading families and craft organisations rather than kings. Standards were enforced because traders needed them, not because a ruler demanded them. For: explains the obsession with weights, seals and measures, and the absence of royal display — merchants show wealth in goods, not monuments. Against: merchant power alone rarely builds and maintains city-wide drains and platforms for centuries.
Authority was held by priests whose power rested on belief rather than violence — which would explain the near-absence of weapons. For: would account for compliance without an army. Against: no temples have been identified, and no religious texts survive. This is largely an argument from absence.
Different groups held different kinds of authority, none of them supreme — one body over water, another over trade, another over ritual, cooperating without a pyramid of command. For: fits the evidence better than any of the others, because it predicts exactly what we see: strong coordination with no visible ruler. Against: it is very hard to prove, and risks becoming a way of saying "we do not know" in more impressive language.
What would settle it: reading the script would probably do it in an afternoon. Short of that: excavating one of the barely-touched large sites, especially Ganweriwala, or finding a building that is unambiguously a palace or a temple.
The genuinely interesting possibility: that the question itself is badly framed. "Who was in charge?" assumes somebody was. If the Harappans ran a complex urban society without a single ruler, they did something no other early civilisation managed, and our categories are simply the wrong shape for them.
Notice what makes the Harappan case feel so strange: we assume complexity requires hierarchy. Big cities need a big boss.
Where does that belief come from? From the other early civilisations we know about. Egypt had pharaohs. Mesopotamia had kings. China had emperors. So: cities need kings.
But that is reasoning from a sample of three or four. It is like meeting four tall people from one country and concluding everyone there is tall.
Harappa may simply be the counter-example. If so, the honest response is not to keep hunting for the palace we have failed to find for a hundred years. It is to consider that our idea of how big societies must work was built on too few cases.
I want to be even-handed here: the opposite error is just as easy. Wanting Harappa to be a peaceful, egalitarian, kingless utopia is a temptation too, and some popular writing gives in to it completely. Absence of evidence for kings is not evidence of absence. Wooden palaces rot. Excavation is patchy. Perhaps five per cent of the total area has been dug.
The correct position is uncomfortable and correct: we do not know, and both stories remain possible.
4.5 — Was it really peaceful?
The claim that Harappan society was unusually peaceful is popular and it is not baseless. But it needs handling carefully.
What supports it: very few weapons compared with contemporary Egypt or Mesopotamia; no armour; no clearly military buildings; no art showing warfare, captives or victory; no destruction layers at most sites; no mass graves of battle dead.
What complicates it: the cities do have walls. Large ones. They are usually explained as flood defences and as boundaries controlling who and what came in — which is plausible on a flood plain — but a wall is a wall. Some skeletons show head injuries. And copper tools can be used on people; the line between an axe and a weapon is about intention, which does not fossilise.
The fair summary: Harappan society looks markedly less militarised than its contemporaries. That is a real and interesting finding. It is not the same as proving it was peaceful, and it says nothing about whether it was internally fair.
Harappa left no palaces, no royal tombs, no royal art, no boastful inscriptions and not one person's name.
The famous "Priest-King" statue is a guess wearing a label. There is no evidence he was either.
Yet the weights are precisely standardised across hundreds of kilometres — hard physical proof that something was organising things.
Five theories compete: one empire, many city-states, merchant guilds, priests, or no single top at all. Nobody knows.
The most interesting possibility is that "who was in charge?" is the wrong question, and Harappa is the counter-example to our belief that big societies need a boss.
5The Writing Nobody Can Read
Around four thousand inscriptions survive. A century of effort by linguists, archaeologists, code-breakers and computers has produced not one accepted reading. Some researchers now argue it is not writing at all.
5.1 — What the inscriptions are on
Most Harappan writing appears on seals. These are small squares of soapstone, usually two to four centimetres across, carved with a picture and a short line of signs, then fired to harden the surface. On the back is a small pierced knob so the seal could be hung on a cord.
A seal is a stamp. You press it into soft clay and it leaves an impression. Impressions have been found on lumps of clay that were once wrapped around the fastenings of bundles and pots — so the seals were used to mark ownership, to certify a shipment, or to seal a container so tampering would show.
Steatite (soapstone): a soft stone easy to carve with simple tools, which hardens when heated. Ideal for making small detailed objects.
Script: a system of written signs. Language: the speech the signs record. These are different things — the same script can write several languages, and the same language can be written in several scripts. This distinction matters enormously in this chapter.
The pictures on the seals are mostly animals. By far the most common is a single-horned creature facing a small object often called a "standard" or "offering stand" — usually nicknamed the unicorn. It appears on more seals than every other motif combined. There are also bulls, elephants, rhinoceroses, tigers, buffalo, and a few complicated scenes with human or part-human figures.
Writing also appears on copper tablets, pottery, ivory rods, and a handful of larger objects — including the famous Dholavira signboard, a row of ten large signs made of inlaid material, found fallen near a gateway. It seems to have been mounted above the gate, which would make it something like a public sign — the only one of its kind ever found.
5.2 — Why it has resisted for a century
Codes get broken all the time. Egyptian hieroglyphs were cracked. Linear B, the script of Bronze Age Greece, was cracked. Why not this one?
Three reasons, and together they are close to fatal.
Reason one: the inscriptions are extremely short. The average is about five signs. The longest on a single surface is around seventeen. Compare that with a Mesopotamian clay tablet carrying hundreds of words, or an Egyptian tomb wall covered in continuous text.
Try to work out an unknown language from a collection of five-word fragments with no context.
Now imagine all you have are the labels off packaging. "Best before." "Made in." "Contains nuts." Thousands of them, in a language you have never heard, with no grammar, no sentences, and nothing telling you what any product is.
That is roughly the situation. The corpus is large in number and tiny in content.
Reason two: there is no bilingual text. Egyptian hieroglyphs were broken because of the Rosetta Stone, which carried the same message in Greek — a language scholars already knew. Nothing of that kind exists for the Harappan script. A few Indus seals turn up in Mesopotamia, but with no translation attached.
Reason three: we do not know the language underneath. This is the worst one. With Linear B, Michael Ventris eventually guessed the language was an early form of Greek, and once he tried that, everything fell into place. For the Harappan script we do not know whether the language was an early Dravidian language, an early Indo-Aryan one, something related to the Munda languages, or a language that has left no descendants at all and is simply gone.
Unknown script plus unknown language plus no long texts plus no bilingual is, as things stand, an unsolvable combination.
5.3 — Is it even writing?
In 2004 a group of researchers published a paper arguing that the "script" does not record language at all — that the signs are non-linguistic symbols, like heraldic badges, ownership marks, religious emblems or trade tokens. Illiterate societies use complex symbol systems all the time.
The texts are too short. No known writing system has an average text length of five signs. Real writing produces long texts because people write down long things.
The sign count is awkward. Roughly four hundred signs, depending on how you classify them — too many for an alphabet, too few for a Chinese-style system where each sign is a word.
It never got longer. Writing systems grow with use. This one shows no development towards continuous text in seven hundred years.
It vanished completely. Real writing tends to survive its civilisation and get adapted. This disappeared without descendants — easier to explain if literacy was never widespread because there was nothing to be literate in.
Statistical structure. Studies of sign order show the sequences behave like language: certain signs strongly prefer certain positions, some sign pairs are common and others never occur. Measures of predictability place the Indus signs close to known written languages and away from non-linguistic sign sets. The signs are ordered by rules.
Short is explainable. If most surviving inscriptions are on seals — that is, on labels and stamps — short is exactly what you expect. Longer texts might have been written on cloth, palm leaf, bark or wood, none of which survive in this climate.
Direction is consistent. Signs appear to run right to left, shown by cramping at the left end where a carver ran out of room. Consistent direction is a hallmark of writing.
The Dholavira signboard. Ten large signs, publicly displayed above a gate. That is a strange thing to do with random symbols.
What would settle it: one long text. A single inscription of a hundred signs would end the argument immediately. The best hope is an unexcavated site — or waterlogged ground, where organic material survives.
Why the statistical argument is not conclusive: other rule-governed non-linguistic systems, like heraldry or ritual sequences, could in principle produce similar patterns. The tests show the signs follow rules. They cannot yet show the rules are grammar.
5.4 — And if it is writing, what language?
The most widely supported proposal in academic work. Dravidian languages — Tamil, Telugu, Kannada, Malayalam and others — are today spoken mainly in southern India, but Brahui is still spoken in Balochistan, within the old Harappan zone, which may be a surviving pocket. Some structural features of the script have been argued to fit Dravidian grammar. Weakness: no proposed reading has been independently verified, and Brahui's presence in the north-west can be explained by later migration.
Would make the Harappans the same people who later composed the Vedas, and Indian civilisation continuous from Harappa to the present. Weakness, and it is severe: the Rigveda describes a society with horses and spoked-wheel chariots at its centre. Neither is securely attested in mature Harappan remains. Proposed decipherments on this basis have not been accepted, and one high-profile attempt in 2000 was found to rest on a doctored image — see below.
The Munda languages of eastern India are a third family. Or the language may simply be extinct, with no descendants — which happens often. Weakness: unfalsifiable in practice. If the language is gone, we can never confirm a reading.
A necessary caution. In 2000, a book announced a decipherment showing a horse on an Indus seal — which would have been a major result, given the horse's absence. The image turned out to be a computer-altered version of a known seal showing a unicorn, with the horn removed and the outline reshaped. It was exposed publicly and quickly.
I include this not to score a point but as a warning that runs in every direction. When people badly want the past to be a particular shape, the temptation to help the evidence along is real. Watch for it in others. Watch for it in yourself. And notice which claims about Harappa you find yourself hoping are true — that is where you are least reliable.
About four thousand inscriptions survive, mostly on stamp seals. The average is around five signs. The longest on one surface is about seventeen.
It cannot be read because there is no bilingual text, the inscriptions are far too short, and we do not know the underlying language.
Some researchers argue it is not writing at all. Statistical studies suggest it probably is. Not settled.
The language might be early Dravidian, early Indo-Aryan, Munda, or something extinct. Nobody knows.
One long inscription would end a century of argument in a single afternoon. It has not been found.
6Work, Food and Trade
Cotton four thousand years before the mills of Manchester. Beads drilled with a precision that modern craftsmen struggle to match. Ships to Mesopotamia. And one animal, conspicuously missing, that will start a war in Part 3.
6.1 — What they grew
The foundation was wheat and barley, sown in the cool season and harvested in spring. Alongside them: peas, lentils, chickpeas, sesame, mustard, dates and melons. Millets and rice appear increasingly in the later period, especially in the east and south.
And cotton. The Harappans were growing and weaving cotton, and the earliest cotton fibres in the world come from this region. Actual scraps of cotton cloth have survived at Mohenjo-daro, preserved by contact with corroding metal.
Cotton is the fabric that later built and broke empires. Britain's industrial revolution ran on it. India's weavers were ruined by it. The American slave economy grew it.
It starts here. Four and a half thousand years ago, in and around Punjab, people were spinning and weaving a plant fibre that no other civilisation of that era used. Egypt wore linen. Mesopotamia wore wool. The Harappans wore cotton.
6.2 — What they kept
Bones from rubbish heaps tell us the animals: zebu cattle (the humped Indian breed, domesticated locally, as Part 1 covered), water buffalo, sheep, goats, pigs, chickens, dogs, cats. Elephants appear on seals and were probably tamed if not fully domesticated. Camels appear late.
Cattle dominate. They pulled carts and ploughs, gave milk, and appear constantly in the art. Whatever the Harappans thought about the world, cattle were near the centre of it.
6.3 — The animal that is not there
Now something that matters far beyond this chapter.
There is no clear evidence of the domestic horse in mature Harappan sites.
Not on a single seal — and the Harappans depicted bulls, buffalo, elephants, rhinos, tigers, crocodiles and antelope in loving detail. Not in the bone assemblages in any way that specialists agree on. No horse gear. No spoked wheels, which are the technology that makes a horse-drawn chariot possible.
Some excavators have reported horse bones at Harappan sites, most notably at Surkotada in Gujarat, and a specialist in horse anatomy examined those bones and identified them as domestic horse. There are also scattered claims from other sites. If correct, the horse was present, just rare.
Distinguishing horse bones from those of the wild ass, the onager and the half-ass is genuinely difficult, and specialists who re-examined the Surkotada material disagreed with the identification. Beyond the disputed bones there is nothing: no horse in the art, no horse gear, no spoked wheels, no dung, no burials. For an animal that Vedic and later societies celebrated constantly, the silence is total. A handful of contested bones is not much against that.
Why anyone cares this much. The Rigveda is soaked in horses and chariots. If the Harappans had no horses, then the Harappans and the Vedic people look like different populations at different times — which supports the idea that Indo-Aryan speakers arrived after the cities declined. If the Harappans did have horses, that gap narrows.
So this is not a debate about bones. It is the front line of the biggest argument in Indian history, which is Part 3's subject. I am flagging it here so that when you meet it there, you already know the horse question is load-bearing and you can watch how each side handles it.
6.4 — The craftsmen
Harappan craft work is where this civilisation is at its most impressive, and it is the part that gets the least attention.
Carnelian beads. Carnelian is a hard orange-red stone. The Harappans made long tube-shaped beads from it — some over ten centimetres — with a hole drilled straight down the middle. Drilling a narrow, perfectly straight hole through ten centimetres of a stone harder than steel, with no metal drill bits, is extraordinarily difficult. They did it using specially made drill heads of an unusually hard rock, turned with a bow drill, with abrasive and water, over days of continuous work for a single bead. The main production centre was Chanhudaro.
Shell bangles. Cut from thick conch shell brought from the coast, sawn into rings and polished. Found across the whole region, far inland.
Faience. A glassy ceramic made from crushed quartz, coloured and fired. Used for beads, small figures and tablets.
Copper and bronze. Tools, vessels, mirrors, razors, fish-hooks, and figures cast by the lost-wax method — including the small bronze often called the Dancing Girl, about ten centimetres tall, standing with one hand on her hip and an arm stacked with bangles.
How do you prove how a bead was drilled? You look inside the hole with a microscope. The drilling leaves fine spiral scratch marks whose spacing and depth indicate the drill type and the abrasive used.
Then you do experimental archaeology — you make the tools yourself, from the same materials, and try to reproduce the bead. Modern researchers doing this have found that a single long carnelian bead takes many days of continuous drilling. That number is not a guess from the object; it is a measured result from someone actually doing it.
This also tells us something social. A bead that takes over a week of a skilled worker's time is not something everybody owned. Craft time is a form of wealth, and it is one of the few ways we can detect inequality in a society that left no palaces.
6.5 — Trade, and a name in a foreign language
Harappan goods travelled a very long way. Indus seals, carnelian beads and weights have been found in Mesopotamia — in the cities of what is now Iraq. Trade went by sea along the coast of the Arabian Sea and Persian Gulf, and overland through Iran and Afghanistan.
Coming the other way: lapis lazuli from Afghanistan, turquoise from Iran, gold and silver, copper from Rajasthan and Oman.
And here is one of the most tantalising facts in the whole subject.
Mesopotamian records — which we can read — mention trading partners to the east. One is called Meluhha. The texts describe Meluhha as a source of carnelian, ivory, exotic woods and animals, and refer to Meluhhan ships, a Meluhhan village in Mesopotamia, and an official who worked as a Meluhhan interpreter.
Almost everyone agrees Meluhha is the Harappan world.
Think about what that means, and how strange it is.
We know what the neighbours called them. We do not know what they called themselves.
We have foreign records describing their ships, their goods and even a man employed to translate their language — and not one word of the language he translated.
That interpreter could have solved this entire chapter for us. He is four thousand years out of reach.
6.6 — Ordinary life
The small finds are, in some ways, the most human part of the record.
Children's toys: little clay carts with working wheels, whistles shaped like birds, model animals with heads that nod on a string, tiny cooking pots. Gaming pieces and dice — cubical dice with marked faces, essentially identical to modern ones. Combs, mirrors, razors, ear-cleaners, cosmetics.
Skeletal studies suggest a generally reasonable diet with the usual ancient problems — worn teeth from grit in stone-ground flour, arthritis, healed fractures. There is some evidence of differences in health between burial groups, which hints at inequality, but the sample is small and the conclusions are cautious.
Wheat, barley, pulses, sesame — and cotton, the world's earliest, grown in and around Punjab.
No clear evidence of the horse in mature Harappan sites. Remember this. It becomes the front line of the biggest argument in Indian history in Part 3.
Craft work was superb: carnelian beads taking over a week each to drill, shell bangles, faience, lost-wax bronze.
They traded with Mesopotamia, who called them Meluhha and even employed an interpreter for their language.
We know what foreigners called them. We do not know what they called themselves.
7What Did They Believe?
No temples. No gods anyone can name. No prayers, because we cannot read. Everything in this chapter is guesswork built on a handful of objects — and this is the chapter where guesswork gets least honest.
7.1 — The size of the problem
For Egypt, religion is easy. There are temples, painted tomb walls, named gods with named jobs, and mountains of readable text explaining exactly what happens after you die.
For Harappa there is: some small clay figures, some seals with strange scenes, some tanks and platforms of unknown purpose, and burials.
That is all. Everything you have ever read about Harappan religion is interpretation of those four things.
7.2 — The seal called Pashupati
One seal from Mohenjo-daro is more famous than all the others put together.
It shows a figure seated in a posture with the legs bent and the heels together. The figure has an elaborate horned headdress. Around it are animals — an elephant, a tiger, a rhinoceros, a buffalo, and deer below.
John Marshall called it a proto-Shiva, and it is widely known as the Pashupati seal — "lord of animals", a title of Shiva.
The argument for the identification is a set of resemblances to the later god Shiva: seated in something like a yoga posture, surrounded by animals as a lord of beasts, with what some read as three faces and possibly an erect phallus. Shiva is associated with all of these.
The resemblances are several and specific, not one loose similarity. Yogic posture, lordship over animals, possible three faces, possible horns, all matching a later god who has exactly those attributes. Shiva is often argued to be a non-Vedic deity absorbed into the mainstream later, which would fit a pre-Vedic origin. And there is real cultural continuity in this region across long periods.
The posture may be a sitting position, not yoga. Similar seated figures appear in Mesopotamian and Elamite art with no yogic meaning.
"Lord of animals" is a near-universal motif. Figures flanked by beasts appear across the ancient world from Mesopotamia to Europe. It is one of the commonest images in early art, and its presence proves nothing about a specific god.
The three faces are not clear. On an object this small and worn, what looks like faces may be a headdress or damage. Some scholars see a buffalo mask.
It might not be a god. It could be a person, a chieftain, a mythical hero, a ritual performer in a mask, or a scene from a story we will never know.
Above all: this is reading backwards. We are matching an object from 2500 BCE against a god described in texts written a thousand or more years later, using the later text to interpret the earlier object.
The fair position: it may show a deity. That deity may have contributed something to what later became Shiva. Neither can be shown, and the confident version you will meet online is not supported by what is actually on the stone.
The whole chapter turns on one habit of mind, so let me name it as clearly as I can.
Reading backwards: using a later, well-documented thing to explain an earlier, undocumented thing, because the later thing is the only guide you have.
It is not stupid. Sometimes it is right — cultures do keep things for a long time. But it has a built-in failure that is very hard to see from the inside: it can only ever find what you already know to look for.
If you approach the Harappans holding Hinduism in your hand, you will find bits of Hinduism, because you will recognise those bits and pass over everything else as unidentifiable. The parts of their religion that survived into later times will glow. The parts that died with them — which may have been most of it — will be invisible.
Turn it round and see how strange it is. Suppose someone three thousand years from now finds a cross-shaped ornament in the rubble of a modern Indian city and concludes the whole society was Christian. The object is real. The reasoning is not.
The safest reading of the Harappan evidence is that we are looking at a religion we do not recognise and have no name for. That is unsatisfying. It is also probably true.
7.3 — The figurines
Harappan sites produce large numbers of small clay figures, many of them female, often with elaborate headdresses and jewellery. These are traditionally called "mother goddesses" and read as evidence of a fertility cult.
The same caution applies, and here the counter-argument is stronger than most people realise.
Points against the goddess reading: the figures are found in ordinary domestic rubbish, not in special deposits or shrines. They are crudely made. There are male figures, animal figures, and figures that are hard to sex, in the same contexts. And — this is the one that gives people pause — a small clay figure of a person found in a house might simply be a doll. Children have always had them.
Points for it: there are a lot of them, some are elaborately decorated, and some have smoke-blackening on the head which has been read as a small lamp or offering. Household religion is real and normally leaves exactly this kind of trace — modest objects in ordinary homes.
Fair summary: some were probably religious. Some were probably toys. Some were probably both, since that distinction may not have existed for them. Calling the whole category "mother goddesses" decides the question by naming it — the same trap as the Priest-King.
7.4 — The burials
Harappan burial is usually a body laid out flat on its back, in a simple grave pit, oriented roughly north-south, with a few pots and occasionally personal ornaments.
What matters is what is not there. No spectacular tombs. No mass of grave wealth. No sacrificed servants or animals. No pyramid, no ziggurat, no royal necropolis.
Compare the tomb of Tutankhamun — a minor pharaoh who died young, buried with a solid gold coffin, a gold mask and thousands of objects — with the burials of a civilisation larger than Egypt, which contain some pots.
Two readings are possible, and both are respectable. Either Harappan society was genuinely less unequal in death than its contemporaries. Or its elites were disposed of some other way — cremation, river disposal, exposure — that leaves nothing to dig up. Both are live. Note that later South Asian practice leans heavily towards cremation, which would leave almost no trace at all.
7.5 — Fire altars, and a word about wanting
At Kalibangan, and at a few other sites, excavators found rows of brick-lined pits containing ash, charcoal and small clay objects. These have been described as fire altars and connected to later Vedic fire ritual.
If that is right, it is a significant link between Harappan and Vedic practice.
The objection is straightforward: a pit with ash in it is a pit with ash in it. Cooking hearths, industrial hearths, rubbish burning and ritual fire can all leave similar traces, and telling them apart requires evidence — residue analysis, the position of the pits, what is around them — rather than resemblance.
This one is genuinely undecided, and it deserves better excavation rather than louder argument.
No temples have been identified. No god can be named. Nothing can be read. Everything about Harappan religion is interpretation of a few objects.
The famous Pashupati seal may or may not show an early form of Shiva. The evidence is much weaker than popular accounts suggest.
The "mother goddess" figurines come from household rubbish. Some may be religious. Some may be dolls.
Burials are modest and fairly equal — either a real feature of the society, or a sign that elites were cremated and left nothing behind.
The biggest danger in this chapter is reading backwards: using later Hinduism to explain earlier objects, and so only ever finding what you already knew.
8The Saraswati Question
In Part 1 I promised to come back to the river that vanished. Here it is: a dry channel, an ancient hymn, hundreds of settlements, some difficult geology, and one of the most politically loaded arguments in Indian history.
8.1 — The facts everyone agrees on
Start with what is not in dispute. There is quite a lot of it.
One. There is a dry riverbed. Running roughly parallel to the Indus, east of it, through Haryana, Rajasthan and Cholistan in Pakistan, there is the visible bed of a large ancient river. It is called the Ghaggar in India and the Hakra in Pakistan. Today it carries only seasonal water and dies in the desert. Satellite imagery shows the old channel clearly, in places several kilometres wide.
Two. Harappan sites cluster along it — heavily. A very large proportion of known Harappan settlements sit along this dry channel rather than along the Indus. Ganweriwala, one of the largest sites, is on it. Kalibangan and Banawali are on it. Rakhigarhi is in its wider system.
Three. The Rigveda describes a river called the Saraswati. It is praised in the highest terms — as great among rivers, as flowing from the mountain to the sea. It is named alongside other rivers in a sequence that places it between the Yamuna and the Sutlej, which is roughly where the Ghaggar-Hakra is.
Four. Later texts describe the Saraswati as disappearing. The Mahabharata and other later works refer to the river vanishing into the sands at a place called Vinashana.
Those four facts, put side by side, make an obvious story: the Saraswati was the Ghaggar-Hakra, it was mighty, the Harappans lived along it, it dried up, and that is why the cities died.
It is a good story. The trouble is in the timing.
8.2 — What the geology says
Palaeochannel: the buried or dried-up bed of a river that no longer flows there. From Greek palaios, "ancient".
Provenance analysis: working out where sand and sediment originally came from, by examining the tiny mineral grains in it. Different mountains produce different mineral signatures, so a grain of sand carries a fingerprint of its home. This lets geologists tell whether a river was fed by Himalayan glaciers or only by local rain.
Luminescence dating: a method that measures how long a grain of sand has been buried away from sunlight, giving a date for when the sediment was laid down.
Geologists have drilled the Ghaggar-Hakra channel, dated the sediments, and traced where the sand came from. Several major studies over the past fifteen years have converged on a picture, and it does not fit the obvious story.
The findings, in plain terms:
- The channel was once fed by a large Himalayan river. That much is confirmed — the mineral fingerprints in the deeper sediments point to Himalayan sources, carried by the ancestral Sutlej and possibly the Yamuna.
- But those connections were lost long before the Harappan cities. The dating suggests the Sutlej swung away from this channel thousands of years earlier — estimates of several thousand years before the mature Harappan period.
- During the mature Harappan period itself, the evidence suggests the Ghaggar-Hakra was a monsoon-fed seasonal river, not a glacier-fed perennial one. Substantial in a good year. Not a giant.
The channel is enormously wide in places, which suggests a large flow. The Rigveda's praise is specific and emphatic. The sheer density of large settlements along it implies a reliable water supply capable of supporting cities. Some dating results are contested and the sampling is patchy — a handful of cores cannot fully characterise a river system a thousand kilometres long. And oral tradition can preserve real memory of a landscape over long periods.
Sediment provenance is a physical measurement, not an interpretation: the mineral grains from the Harappan-period layers do not carry the Himalayan glacial signature. Multiple independent studies using different methods have reached compatible conclusions. A wide channel records the river's past size, not its size at any one moment — a large old bed can carry a small modern stream, which is exactly what the Ghaggar does now.
And there is a striking twist: a monsoon-fed river may be better for settlement than a glacier-fed one. Himalayan rivers flood violently and change course. A steadier seasonal river is more predictable and safer to build beside — which would explain why so many towns are here rather than on the wild Indus.
What is agreed by everyone: a substantial river system did run here; large numbers of Harappans lived along it; and it dried up. The dispute is about how big it was during the Harappan period and whether it was glacier-fed.
What would settle it: far more dated cores along the whole length of the channel, especially in Pakistan, where far less work has been done.
8.3 — Why this argument is so heated
You may be wondering why a question about sediment dating produces such anger.
It is because of what it is attached to.
If the Saraswati of the Rigveda was a mighty river during the Harappan period, then the people who composed the Rigveda were plausibly there, at that time, watching that river. That would make the Harappans and the Vedic people the same population — and Indian civilisation continuous and home-grown, with no significant arrival from outside.
If the Ghaggar-Hakra was already a modest seasonal river by then, the Rigvedic descriptions either refer to a much older memory, or to a different river, or are poetic praise rather than geography. That leaves room for Indo-Aryan speakers arriving later, from outside — which is the mainstream position and is politically unwelcome to many.
So the name "Indus-Saraswati Civilisation" from Chapter 1 is not a neutral label. It carries this whole argument inside it.
Two assumptions run underneath this debate, and almost nobody states them.
First: that a name in a text must refer to the thing we now attach that name to. River names move. People carry names with them and reuse them for new places — the world is full of new Yorks, new Delhis and new Amsterdams. A hymn praising a Saraswati is not automatically praising this channel.
Second: that praise poetry is geography. The Rigveda is religious poetry, not a survey report. When a hymn calls a river greatest among rivers and says it flows from mountain to sea, that may be description — or it may be the ordinary language of devotion. We do not read "my love is a red rose" as botany.
Neither assumption is necessarily wrong. Both are usually invisible, and both do enormous work in the argument.
And one more, aimed at everyone including me: the question "were the Harappans the ancestors of today's Indians?" often means "are we the same as them?" — and that is not really an archaeological question. Populations mix constantly. Ancient DNA shows that almost everyone in South Asia today carries ancestry from several ancient sources, including Harappan-related ancestry. The honest answer to "are they us?" is: partly, along with several other groups, and this is true of nearly every population on Earth.
A large river once ran east of the Indus, through what is now Haryana, Rajasthan and Cholistan. Its dry bed is the Ghaggar-Hakra, and huge numbers of Harappan sites sit along it.
The Rigveda praises a great river called the Saraswati in roughly that position, and later texts say it vanished into the sand.
Geology says the Himalayan feed was cut off long before the Harappan cities, and that during Harappan times it was a monsoon-fed seasonal river — possibly a better neighbour than a violent glacier-fed one.
The argument is fierce because it decides whether the Harappans and the Vedic people were the same population.
Everyone agrees a river ran there and dried up. The fight is over how big it was, and when.
9Why Did It End?
Six explanations have been offered. One was believed for decades and turned out to be badly wrong. The current answer is less dramatic than any of them, and it changes what the question should have been.
9.1 — What actually happened
First, be precise about what needs explaining, because most accounts get this wrong from the start.
Between roughly 1900 and 1700 BCE, over several generations:
- The big cities lost population and were gradually abandoned.
- Municipal upkeep stopped — drains blocked, streets were built over, standards of construction fell.
- The weight system stopped being used.
- The script disappeared.
- Long-distance trade with Mesopotamia fell away.
- Seals stopped being made.
But note carefully what did not happen:
- The population did not disappear. Settlements continue in large numbers, especially eastward, towards the Yamuna and the upper Ganga plain.
- Farming continued. Crops continued. Pottery traditions continued in modified form.
- There is no layer of burning and destruction across the sites.
- There are no mass graves of war dead.
So this is the end of cities and of a shared standard system — not the end of a people. That distinction runs through everything below.
9.2 — The theory that was wrong
For much of the twentieth century the standard explanation was invasion.
The archaeologist Mortimer Wheeler found groups of skeletons in the upper levels at Mohenjo-daro, some apparently lying where they fell. He connected them with passages in the Rigveda in which the god Indra destroys fortified places, and concluded that Aryan invaders had stormed the cities. He wrote that Indra stood accused.
It is a vivid image and it went into textbooks worldwide. It is now rejected by essentially all specialists. Here is why.
The skeletons were re-examined in detail by later researchers, and the case fell apart on several independent grounds.
They are not contemporaneous. The bodies come from different levels and different periods. They were not killed together in one event.
Most show no injuries. A massacre leaves cut marks, crushing injuries, arrow damage. Most of these skeletons show none.
Some had been buried properly. Not left where they fell.
The city was largely empty already. The layers show decline before the period the skeletons come from. There was no thriving city to storm.
And there is nothing else. No destruction layer. No weapons in the streets. No burnt gates. No invader settlement built on top. A conquest leaves a great deal of physical evidence, and none of it is there.
This is a model case of how the discipline is supposed to work. A dramatic claim, made by a respected figure, matching the political assumptions of its time, examined properly, and abandoned. Keep it in mind for Part 3, where the descendants of this argument are still being fought over.
9.3 — The theories still standing
Records from lake sediments, cave deposits and ocean cores show a significant weakening of the monsoon across a wide region in the centuries around 2200–1900 BCE. Less rain means smaller harvests, and a city depends absolutely on a food surplus from its countryside. Strength: the timing lines up well, the evidence is independent of archaeology, and the same climatic event coincides with severe trouble in Egypt and Mesopotamia. This is currently the leading factor. Weakness: people adapt to drought; it explains pressure better than it explains the loss of writing and standards.
As Part 1 explained, rivers on flat land shift course. If the water moved away from a city, the city was finished regardless of anything else. The drying of the Ghaggar-Hakra system fits here. Strength: mechanically certain to kill individual settlements, and there is direct evidence of channel change. Weakness: hard to explain a simultaneous region-wide collapse, since rivers shift at different times in different places.
Mesopotamia went through its own upheavals in this period, and references to Meluhha fade from its records. If a significant part of the urban economy rested on long-distance trade, losing that market would hit the cities hardest and the villages least. Strength: would explain why cities specifically failed while rural life continued. Weakness: probably not big enough on its own to bring down a whole civilisation.
Some skeletal evidence has been read as showing increased infectious disease in the later period. Dense cities with shared water are vulnerable. Strength: plausible; cities are disease amplifiers. Weakness: the evidence is thin and the samples are small. This is a possibility, not a finding.
The cities ended, but the people did not. Settlement shifted east into better-watered land towards the Ganga-Yamuna region, and society reorganised into smaller units. On this reading nothing was destroyed; a particular way of living stopped being worth the effort. Strength: matches the archaeology best. Late Harappan settlements are numerous, and the population appears to move rather than vanish.
What would sharpen it: more precisely dated local climate records, and more excavation of Late Harappan sites, which have been badly neglected because they are less spectacular.
The reframing that matters: the question should not be "what killed them?" It should be "why did people stop finding it worth living in cities?" Cities are expensive. They need surplus, coordination and maintenance. When the return falls below the cost, people leave — and that is not a catastrophe, it is a decision, made by many people separately over a century or two.
9.4 — What the DNA says
One more strand, because it is recent and it matters.
Ancient DNA is difficult to recover in hot climates, where it degrades fast. Despite that, a genome has been sequenced from a Harappan-period individual buried at Rakhigarhi in Haryana.
The results, stated carefully:
- The ancestry is a mixture of an Iranian-related component — from a lineage that separated early, before farming began in the Near East — and an indigenous South Asian hunter-gatherer component.
- No Steppe ancestry was found. Steppe ancestry is associated with Indo-European-speaking populations from Central Asia and is widely present in South Asia today.
- No Anatolian farmer ancestry was found, which argues against farming having been brought into the region by a large migration of Near Eastern farmers.
What this supports: the Harappans were largely local in origin, and Steppe-related ancestry entered South Asia mainly after the Harappan period. Both of those are relevant to Part 3.
Ancient DNA results get quoted with far more confidence than they deserve, in every direction. Two cautions.
One genome is one person. A single individual cannot represent a civilisation of a million-plus people spread over 1.5 million square kilometres. It would be like sequencing one person from Chandigarh and reporting the genetics of India.
Genes are not language and not culture. This is the big one. People change language without changing ancestry all the time — most of India speaks a language its ancestors did not speak two thousand years ago, and English is now spoken by hundreds of millions with no English ancestry whatsoever. Finding or not finding a genetic signature does not directly answer a question about who spoke what.
The result above is real, careful work and genuinely informative. It is also one skeleton, and it cannot bear the weight that public argument piles onto it.
Between about 1900 and 1700 BCE the cities emptied, standards broke down, and the script disappeared — but the people did not vanish. They moved east and lived in smaller settlements.
The old invasion theory is dead. The "massacre" skeletons at Mohenjo-daro were from different periods, mostly uninjured, some properly buried, in a city that was already emptying.
The current view is a combination: a drying climate, shifting rivers, and contracting trade made a costly urban system no longer worth maintaining.
DNA from Rakhigarhi shows local ancestry with no Steppe component — but it is one individual, and genes are not language.
The better question is not "what killed them?" but "why did cities stop being worth it?"
10An Honest List of What We Do Not Know
This part contains more uncertainty than Part 1 did. Here is exactly where it sits.
Genuinely unknown
- What they wrote. Not one sign has an agreed meaning.
- What language they spoke. Dravidian, Indo-Aryan, Munda, or something extinct.
- Whether the script is writing at all. Probably, not certainly.
- Who governed, and how. Five competing models, no decisive evidence.
- What they believed. No temple, no god, no readable prayer.
- What they called themselves. We only have the Mesopotamian name, Meluhha.
- Whether the "citadel", the "granary", the "Great Bath" and the "priest-king" are anything like what those names claim.
- How unequal the society was. Modest burials point one way; craft time and house sizes hint another.
- The exact size and behaviour of the Ghaggar-Hakra during the Harappan period.
- Why cities were abandoned — the combination is agreed, the weighting of each factor is not.
- What is in the ninety-five per cent of the area that has never been excavated, including nearly all of Ganweriwala.
Solid
- The civilisation existed, covered roughly 1.5 million square kilometres, and peaked between about 2600 and 1900 BCE.
- Cities were planned on grids and aligned to the compass.
- Bricks followed a 1:2:4 ratio across the whole region.
- Covered street drains connected to bathing floors in ordinary houses.
- Weights were standardised in a doubling series and are consistent across hundreds of kilometres. This is a physical measurement.
- They grew and wove cotton — the earliest known anywhere.
- They traded with Mesopotamia, who called them Meluhha.
- Around four thousand short inscriptions survive, undeciphered.
- No palace, royal tomb, royal inscription or ruler's name has ever been found.
- The cities were not destroyed by invasion.
- Late Harappan settlement continued and shifted eastward.
Notice how much longer the first list is than the second.
That is the honest state of the subject, and it is worth saying plainly because you will meet a great deal of writing about Harappa — books, videos, social media, sometimes textbooks — that speaks with total confidence about matters in the first list.
When you meet a confident claim about the Harappans, ask one question: which list is this on?
If it is about bricks, drains, weights, crops or trade, confidence is reasonable. If it is about gods, kings, language, ethnicity or who the modern descendants are, confidence is not available — and someone offering it is telling you about themselves, not about the Harappans.
Sources & further reading — Part 2
Timeline of Part 2
All dates approximate. Researchers disagree over the boundaries between phases.
| When | What happened |
|---|---|
| c. 7000 BCE | Farming established at Mehrgarh. Where Part 1 ended. |
| c. 4000 – 3300 BCE | Hakra ware villages across Punjab, Haryana and Cholistan. |
| c. 3300 – 2600 BCE | Early Harappan. Walled towns, regional pottery, early trade. No script, no cities yet. |
| c. 2800 BCE | Ploughed field at Kalibangan, beneath the later town. |
| c. 2600 BCE | Mature Harappan begins. Cities, grid plans, standard bricks and weights, the script, seals. |
| c. 2500 – 2000 BCE | Peak. Trade with Mesopotamia; Meluhha appears in Sumerian and Akkadian records. |
| c. 2200 BCE | A major drying episode affects a wide belt from the Mediterranean to South Asia. Monsoon weakens. |
| c. 1900 BCE | Mature period ends. Cities begin emptying. Standards, seals and script fall out of use. |
| c. 1900 – 1300 BCE | Late Harappan. Smaller settlements; population shifts east towards the Yamuna and upper Ganga. Bara ware in Punjab. |
| c. 1500 BCE onwards | Composition of the Rigveda begins. Part 3 starts here. |
| 1829 CE | Charles Masson describes the ruins at Harappa without understanding them. |
| 1850s CE | Harappan bricks used as ballast for the Lahore–Multan railway. |
| 1873 CE | Alexander Cunningham publishes an Indus seal and assumes it is foreign. |
| 1921 CE | Daya Ram Sahni begins excavating Harappa. |
| 1922 CE | R. D. Banerji finds the older city beneath Mohenjo-daro. |
| 1924 CE | John Marshall announces the civilisation to the world. |
| 1947 CE | Mortimer Wheeler proposes the invasion theory. Later overturned. |
| 2004 CE | Published challenge to the idea that the Indus signs are writing at all. |
| 2019 CE | First Harappan genome published, from a burial at Rakhigarhi. |
Glossary
Every hard word used in Part 2, in plain English.
| Word | What it means |
|---|---|
| Archaeology | Studying the human past through the physical things people left behind. |
| Bitumen | Natural black tar. Waterproof. Used to seal the Great Bath. |
| Cardinal alignment | Lined up with north, south, east and west rather than randomly. |
| Carnelian | A hard orange-red stone, prized for beads. Very difficult to drill. |
| Citadel | The raised western part of a Harappan city. The name implies a fortress; that is an assumption, not a finding. |
| Faience | A glassy ceramic made from crushed quartz, coloured and fired. |
| Ghaggar-Hakra | The dry river channel east of the Indus. Ghaggar in India, Hakra in Pakistan. Identified by many with the Saraswati. |
| Grid plan | Streets laid out in straight lines crossing at right angles. |
| Heterarchy | A society where different groups hold different kinds of authority and none is supreme. The opposite of a pyramid of command. |
| Lost-wax casting | Making a metal figure by modelling it in wax, covering it in clay, melting the wax out and pouring metal into the space. |
| Luminescence dating | Measuring how long a sand grain has been buried away from sunlight. |
| Meluhha | The Mesopotamian name for the Harappan region. The only name for them recorded by anyone who could write. |
| Palaeochannel | The dried-up or buried bed of a river that no longer flows there. |
| Provenance analysis | Working out where sediment originally came from by examining its mineral grains. |
| Reading backwards | Using a later, well-documented thing to explain an earlier, undocumented one. Tempting, sometimes right, and structurally unable to find anything you did not already expect. |
| Script vs language | A script is the set of written signs. A language is the speech recorded. The same script can write many languages. |
| Seal | A small carved stamp, pressed into wet clay to leave a mark of ownership or authority. |
| Steatite | Soapstone. Soft to carve, hardens when fired. What most seals are made of. |
| Steppe ancestry | A genetic component originating in the grasslands of Central Asia, associated with Indo-European-speaking populations. |
| Tell / mound | An artificial hill formed by centuries of building, collapse and rebuilding on the same spot. |
| Type site | The first or defining site of an archaeological culture, which gives the culture its name. |