The Indus Valley civilisation

Foundations

The largest one, and the one nobody knows

Ask most people to name the ancient civilisations and you get Egypt and Mesopotamia. The third was larger than both put together.

At its height, between roughly 2600 and 1900 BCE, the Indus civilisation covered over a million square kilometres, across what is now Pakistan and northwest India, with more than a thousand known settlements. It was contemporary with the pyramids and with Sumer, and it traded with both.

And it is the one we can barely read, because its script is undeciphered and, unlike Egypt and Mesopotamia, it left nothing that announces who was in charge.

What is there

Start with what has been excavated, because it is remarkable on its own terms.

Planning. The major cities are laid out on approximate cardinal orientations, with main streets and smaller lanes, and with distinct raised and lower areas. This is not organic growth; somebody decided the shape before it was built.

Standardisation, over an enormous area. Fired bricks across the whole civilisation follow a consistent ratio of roughly one to two to four in thickness, width and length. Weights are cubes of chert in a consistent series, and they are precise. The same brick proportions and the same weight system, from Gujarat to Punjab, for seven hundred years.

That is the fact that should make you stop. Maintaining a standard across a million square kilometres for seven centuries requires something. What that something was is exactly what is missing.

Water, on a scale nobody else attempted. This is the civilisation’s signature. Houses had bathing platforms with drains running out to covered channels under the streets, with inspection points for clearing blockages. Mohenjo-daro alone had several hundred wells. No contemporary civilisation invested anything like this proportion of its effort in water and drainage.

Craft and trade. Standardised pottery, carnelian beads drilled with extraordinary skill, cotton textiles, and seals carved with animals and script. Indus seals turn up in Mesopotamia, and Mesopotamian texts refer to a trading partner called Meluhha, generally identified with this region.

What is not there

Now the absences, and the discipline is to take them seriously rather than assuming they will turn up.

No palaces. Excavators have looked for a century. There are large buildings, and none has been securely identified as a ruler’s residence.

No royal tombs. Egypt has the pyramids. Mesopotamia has the royal graves of Ur. Indus burials are comparatively modest, with pottery and ornaments, and without the extreme concentration of wealth in a few graves that marks a monarchy elsewhere.

No unambiguous temples. Buildings have been proposed, and none carries the concentrated religious paraphernalia that identifies a temple in Mesopotamia.

No conspicuous army. There are walls, but they are as consistent with flood defence and controlled access as with siege warfare. There is no cache of standardised weapons, and no art depicting conquest, prisoners or a victorious ruler. In Egypt and Mesopotamia, that imagery is everywhere.

No named rulers. The script cannot be read, and nothing in the material record singles out an individual.

Absence of evidence is not evidence of absence, and this many absences, this consistently, across this much excavation, is itself a finding. Either this society organised power very differently, or it organised it in ways that leave no trace we know how to read.

That is genuinely open, and the honest position is to say so rather than to fill the gap with a king.

The Great Bath

The most-photographed structure is a tank at Mohenjo-daro, about twelve metres by seven and just under two and a half deep, with steps at either end, brick laid in gypsum mortar, a bitumen layer to make it watertight, and a drain to empty it.

We know exactly how it was built and nothing about what it was for. Ritual bathing is the usual interpretation, by analogy with later South Asian practice, and it is an inference across two thousand years of silence. It is worth being clear about which of those two statements is evidence and which is a guess.

Dholavira, and the harder problem

The site that best shows what these people could do is not on a great river at all.

Dholavira sits on an island in the Rann of Kutch, in Gujarat, with no perennial river and very little rain. To live there at all required capturing every drop of two seasonal streams, and the city is essentially a water-management system with a settlement inside it: a series of large reservoirs, some cut directly into rock, with dams, channels and stone-built drains, occupying a substantial fraction of the walled area.

It was inscribed as a UNESCO World Heritage site in 2021. Indian researchers have used ground-penetrating radar there to map buried structures without excavating them, which is worth noting because it is the kind of work that expands what is known without destroying it.

Dholavira makes the point that the Indus civilisation was not one thing repeated. It was a set of solutions to different problems, and the solutions were regional.

The river argument

For a long time the story was simple: a great river, now dry, along which hundreds of settlements lay, which then dried up and ended the civilisation. The Ghaggar-Hakra system in northwest India and Pakistan is that river, and it has attracted a great deal of attention for reasons that go beyond archaeology.

The geology has complicated it in a way that is more interesting than the simple version.

Work published in 2017 traced the history of the palaeo-Sutlej and found that it had avulsed away from the Ghaggar-Hakra valley long before the urban phase, leaving behind an abandoned incised valley. Harappan settlements were then established along that valley, which was fed by monsoon rainfall rather than by Himalayan glacial melt.

The counter-intuitive conclusion is that settlement clustered there because the large river had gone. A monsoon-fed seasonal channel floods less catastrophically than a glacier-fed river in spate, so its floodplain is safer to farm and to build on. Earlier work on the region’s fluvial landscapes reached compatible conclusions about the absence of a large glacier-fed river during the mature phase.

This bears directly on a live public argument about the identification of the Ghaggar-Hakra with a river named in later texts. The scientific question, which is what this lesson can address, is what the water was actually doing and when. That question has an evidence-based answer, and it is the one above.

How it ended, which is not how you were taught

The version in older textbooks is invasion: newcomers arriving and destroying the cities, with skeletons at Mohenjo-daro offered as massacre victims.

That reading has not held. The skeletons are from different levels and contexts and do not represent a single event. The cities were not burned or sacked. And the archaeological pattern is not destruction but dispersal.

After about 1900 BCE, urban centres were progressively abandoned. Standardisation breaks down: the weights, the seals, the brick uniformity. Long-distance trade contracts. Settlement shifts eastward, towards regions with more reliable rainfall, and the number of small sites increases while the large ones empty.

Climate is implicated. An oxygen isotope record from Kotla Dahar, a lake in Haryana, records an abrupt weakening of the summer monsoon around 4,100 years ago, part of a wider event visible in several records around 4,200 years ago.

But the honest account is more careful than “drought killed it”. Work on the region’s agriculture shows Harappan communities occupying several distinct environmental zones with different rainfall and river regimes, growing both winter and summer crops, and adapting cropping strategies to local conditions. A civilisation that diverse does not have a single failure mode.

The best current reading is that weakening monsoon rainfall made the large cities harder to sustain, and populations responded by doing what farmers do: moving to where the water was, and living in smaller settlements. The people did not disappear. The cities did.

The script, and whether there is anything to read

Around four hundred to seven hundred distinct signs are known, and the corpus consists mostly of very short inscriptions, averaging about five signs, largely on seals, tablets and pottery. There is no bilingual text.

Two things make decipherment unusually hard. The inscriptions are short, so statistical patterns are thin. And the sign count sits awkwardly: too many for an alphabet, too few for a fully logographic system like Chinese, which suggests a mixed system but does not pin it down.

The argument about it has two serious sides.

One team applied information-theoretic methods, measuring how strongly each sign constrains the next, and reported that the Indus corpus behaves like linguistic scripts rather than like non-linguistic sign systems such as heraldry or deity symbols.

Another group has argued the opposite: that the brevity of the inscriptions, the absence of any long text, and the pattern of sign use are more consistent with non-linguistic symbols — marks of ownership, religious emblems, political signs — than with writing that encodes speech.

Both positions are held by serious scholars. The question is not settled, and claimed decipherments have appeared regularly for a century without convincing anyone outside the group that produced them.

The rural civilisation

One last correction, and it matters for how the whole thing is pictured.

The cities get the attention, and most people lived in villages. Of over a thousand known settlements, only a handful are large. Recent work on rural sites in northwest India has argued that these places had their own complexity, their own ceramic production and their own relationships with the urban centres, rather than being simple dependencies.

That reframes the civilisation from a few great cities with a hinterland into a dense landscape of settlements of many sizes, of which the cities were the largest nodes. And it fits the ending: a society whose weight was in its villages does not collapse when its cities empty. It redistributes.

Deeper

What standardisation does and does not prove

Uniform brick ratios and weight systems are consistent with several arrangements: centralised administration, a shared guild or craft tradition, or convention maintained by the demands of long-distance trade. Mesopotamian standardisation is accompanied by texts naming the authority that imposed it; Indus standardisation is not. The inference from standards to a state is therefore weaker here than the comparison invites, which is precisely why the absence of palaces matters analytically rather than merely as a curiosity.

The water system in detail

At Mohenjo-daro, household bathing floors are typically the most carefully constructed surfaces in a house, sloped and sealed, discharging through terracotta pipes or brick channels into street drains covered with slabs that could be lifted for cleaning. The scale of well provision is exceptional: Jansen’s survey work documented several hundred wells across the excavated area, many rebuilt upward over time as street levels rose. The investment implied, in labour and in fired brick, is comparable to what other civilisations put into temples and tombs.

Reading the 4.2 ka event locally

The Kotla Dahar record derives oxygen isotope ratios from ostracod and gastropod carbonate in a closed-basin lake, where evaporative enrichment tracks the balance of rainfall and evaporation. A pronounced positive excursion indicates a sharp reduction in monsoon rainfall. Its value lies in being a terrestrial record from within the Harappan zone rather than a marine or distant proxy. Correlating it with settlement change still requires archaeological chronologies of comparable precision, which is the limiting factor.

Why the invasion hypothesis failed

The 1940s reading rested on skeletal remains at Mohenjo-daro interpreted as massacre victims, and on the association of incoming populations with the end of the cities. Re-examination showed the skeletons derive from different stratigraphic levels and contexts, some from formal burials, with no evidence of a single violent episode. Excavation has produced no destruction layers at the major sites. Ancient DNA has since added a further constraint, indicating that steppe-related ancestry arrives in South Asia in the second millennium BCE, after the urban phase had already ended.

The limits of the evidence base

Mohenjo-daro’s lowest levels lie below the water table and are largely unexcavated. Much of the early excavation predates modern recording standards. Site distributions are affected by where survey has been conducted and by modern land use, and comparatively few Harappan sites have been excavated to modern standards. Statements about what the Indus lacked should be read against how much of it remains unexamined.

Frontier

Whether the script encodes language

Resolving this requires either a longer text, a bilingual, or a decisive methodological argument. New inscribed material is found occasionally, and no long text has ever appeared, which is itself part of the non-linguistic argument. Improved statistical methods applied to the growing corpus are the realistic near-term route.

How power was organised

The absence of the usual markers has produced proposals ranging from a merchant oligarchy, to competing elite groups without a single ruler, to a more heterarchical arrangement with authority distributed across institutions. Distinguishing these requires evidence types the Indus record has not yet supplied, and it is the field’s central question.

More ancient genomes

The single Rakhigarhi genome constrains far less than a series would. Systematic sampling across Harappan sites, with methods suited to poor preservation in a warm climate, would test whether the civilisation was genetically homogeneous and how it related to preceding and succeeding populations.

Transcript and shot list

Target runtime about 9 minutes. Built on absence as evidence: catalogue what is there, then catalogue what is missing, and take the absences seriously rather than explaining them away. Shot types: ANIM is rendered from code with Manim, AI is a generated shot from Higgsfield with the shared style reference, REAL is agency imagery with credit, CAM is the narrator.

[00:00] S01 · ANIM · Three areas drawn to scale on a world map: Egypt, Mesopotamia, and the Indus civilisation larger than both combined; a count of over a thousand settlements.

Ask most people to name the ancient civilisations and you get Egypt and Mesopotamia. The third one was bigger than both put together. Between about 2600 and 1900 BCE the Indus civilisation covered over a million square kilometres, across what is now Pakistan and northwest India, with more than a thousand known settlements. It was contemporary with the pyramids. It traded with Mesopotamia. And it is the one we can barely read.

[00:40] S02 · REAL · Aerial or ground view of the excavated grid at Mohenjo-daro (ASI or UNESCO; confirm credit line). S03 · ANIM · Street layout on cardinal orientations; main streets and lanes; raised and lower areas distinguished.

Start with what is there, because it is remarkable on its own terms. The major cities are laid out on approximate cardinal orientations, with main streets and smaller lanes, and distinct raised and lower areas. That is not organic growth. Somebody decided the shape before it was built.

[01:10] S04 · ANIM · Fired bricks in a 1:2:4 ratio shown from sites across Gujarat, Sindh and Punjab; chert weight cubes in a consistent series; a duration bar of 700 years across a million square kilometres.

Then standardisation, over an enormous area. Fired bricks across the whole civilisation follow a consistent ratio, roughly one to two to four. Weights are chert cubes in a consistent series, and they are precise. The same brick proportions and the same weights, from Gujarat to Punjab, for seven hundred years. That is the fact that should make you stop. Holding a standard across a million square kilometres for seven centuries takes something. And what that something was is exactly what is missing.

[01:50] S05 · REAL · A covered street drain with lifting slabs at Mohenjo-daro (research image; confirm credit line). S06 · ANIM · A household bathing platform draining through terracotta pipe into a covered street channel; inspection points; several hundred wells across the city.

And then water, on a scale nobody else attempted. This is the civilisation’s signature. Houses had bathing platforms draining out to covered channels under the streets, with inspection points for clearing blockages. Mohenjo-daro alone had several hundred wells. No contemporary civilisation put anything like this proportion of its effort into water and drainage. In Egypt that effort went into tombs. Here it went into plumbing.

[02:30] S07 · ANIM · Five absences listed in turn and held: no palaces, no royal tombs, no unambiguous temples, no weapon caches or conquest imagery, no named rulers. Beside each, the contemporary Egyptian or Mesopotamian equivalent.

Now the absences, and the discipline is to take them seriously instead of assuming they will turn up. No palaces. Excavators have looked for a century; there are large buildings, and none has been securely identified as a ruler’s residence. No royal tombs. Egypt has the pyramids, Mesopotamia has the royal graves of Ur, and Indus burials are comparatively modest. No unambiguous temples. No conspicuous army: there are walls, but they suit flood defence and controlled access as well as siege, and there is no weapon cache and no art showing conquest, prisoners or a victorious king. In Egypt and Mesopotamia that imagery is everywhere. And no named rulers at all.

[03:20] S08 · ANIM · The phrase “absence of evidence is not evidence of absence” appearing, then qualified: this many absences, this consistently, across this much excavation, is a finding. Two live possibilities left standing side by side.

Absence of evidence is not evidence of absence. But this many absences, this consistently, across this much excavation, is itself a finding. Either this society organised power very differently, or it organised it in ways that leave no trace we know how to read. That is genuinely open, and the honest thing is to say so rather than fill the gap with a king.

[03:50] S09 · REAL · The Great Bath at Mohenjo-daro in raking light (ASI or UNESCO; confirm credit line). S10 · ANIM · Its construction: brick in gypsum mortar, a bitumen waterproofing layer, steps at both ends, a drain; then a divider separating what is known from what is inferred.

The most photographed structure is this tank at Mohenjo-daro. About twelve metres by seven, just under two and a half deep, steps at either end, brick laid in gypsum mortar, a bitumen layer to make it watertight, and a drain to empty it. We know exactly how it was built, and nothing about what it was for. Ritual bathing is the usual interpretation, by analogy with later South Asian practice, across two thousand years of silence. Worth being clear which half of that is evidence and which is a guess.

[04:30] S11 · REAL · Dholavira’s reservoirs cut into rock (UNESCO or ASI; confirm credit line). S12 · ANIM · An island in the Rann of Kutch, no perennial river, low rainfall; two seasonal streams captured; a cascade of reservoirs, dams and stone drains occupying much of the walled area; a GPR survey mapping buried structures without digging.

The site that best shows what these people could do is not on a great river at all. Dholavira sits on an island in the Rann of Kutch, in Gujarat, with no perennial river and very little rain. Living there meant capturing every drop of two seasonal streams, and the city is essentially a water-management system with a settlement inside it: reservoirs, some cut straight into rock, with dams, channels and stone drains taking up a substantial part of the walled area. It became a UNESCO World Heritage site in 2021. And Indian researchers have used ground-penetrating radar there to map what is buried without digging it up. Dholavira makes the point that this was not one thing repeated. It was a set of regional solutions to different problems.

[05:20] S13 · ANIM · The simple story: a great river, hundreds of settlements, the river dries, the civilisation ends. Then the geology: the palaeo-Sutlej avulsing away long before the urban phase, leaving an abandoned incised valley fed by monsoon rain.

Then the river argument, where the geology has made things more interesting than the simple version. The simple story is a great river, now dry, lined with settlements, which dried up and ended everything. Work published in 2017 traced the palaeo-Sutlej and found it had moved away from that valley long before the urban phase, leaving behind an abandoned incised valley fed by monsoon rainfall rather than Himalayan meltwater.

[06:00] S14 · ANIM · The counter-intuitive conclusion: settlement clustered there because the big river had gone; a glacier-fed river in spate flooding catastrophically, against a monsoon-fed channel that is safer to farm beside.

And the conclusion is counter-intuitive. Settlement clustered there because the big river had gone. A monsoon-fed seasonal channel floods less catastrophically than a glacier-fed river in spate, so its floodplain is safer to farm and to build on. There is a live public argument about what to call that river system. The scientific question, which is the one this lesson can answer, is what the water was actually doing and when.

[06:35] S15 · ANIM · The invasion hypothesis: skeletons at Mohenjo-daro as massacre victims; then the correction — different levels, different contexts, no destruction layers, no burning.

Now how it ended, which is not how it used to be taught. The old version is invasion: newcomers destroying the cities, with skeletons at Mohenjo-daro as massacre victims. That has not held. The skeletons come from different levels and contexts and do not represent one event. The cities were not burned or sacked.

[07:05] S16 · ANIM · After 1900 BCE: cities progressively abandoned; weights, seals and brick uniformity breaking down; trade contracting; settlement shifting eastward toward more reliable rainfall; small sites multiplying.

The pattern is not destruction but dispersal. After about 1900 BCE the urban centres were progressively abandoned. Standardisation breaks down: the weights, the seals, the brick uniformity. Long-distance trade contracts. And settlement shifts eastward, towards more reliable rainfall, with the number of small sites rising as the big ones empty.

[07:40] S17 · ANIM · The Kotla Dahar oxygen isotope record showing an abrupt monsoon weakening ~4,100 years ago; then Harappan agriculture across multiple environmental zones, winter and summer cropping, varied strategies; a single-cause arrow crossed out.

Climate is involved. An oxygen isotope record from a lake in Haryana shows an abrupt weakening of the summer monsoon around four thousand one hundred years ago, part of a wider event around forty-two hundred. But the honest account is more careful than drought killed it. These communities occupied several distinct environmental zones with different rainfall and rivers, grew both winter and summer crops, and adapted their strategies locally. A civilisation that varied does not have one failure mode. The best reading is that weaker rainfall made the big cities harder to sustain, and people did what farmers do: moved to where the water was, and lived in smaller places. The people did not disappear. The cities did.

[08:20] S18 · REAL · Indus seals with script and animal motifs (British Museum, ASI or National Museum; confirm credit line). S19 · ANIM · Sign inventory of 400-700, too many for an alphabet and too few for a logographic system; inscriptions averaging five signs; then the two rival positions given equal weight.

Which leaves the script. Four hundred to seven hundred distinct signs, mostly on seals, tablets and pottery, in inscriptions averaging about five signs. No bilingual text anywhere. Two things make it unusually hard: the inscriptions are too short for statistics to bite, and the sign count sits awkwardly between an alphabet and a system like Chinese. And the argument has two serious sides. One team applied information theory, measuring how strongly each sign constrains the next, and reported that the corpus behaves like a linguistic script. Another group argues the opposite: that the brevity, the total absence of long texts, and the pattern of use all fit non-linguistic symbols better — marks of ownership, emblems, political signs. Both positions are held by serious scholars, it is not settled, and claimed decipherments have appeared roughly every few years for a century.

[09:00] S20 · AI into ANIM · A small Harappan village at dusk rather than a city; then the settlement hierarchy shown with a handful of cities atop a dense landscape of small sites; the calendar returns.

One last correction. The cities get the attention, and most people lived in villages. Of over a thousand known settlements, only a handful are large, and recent work argues the rural sites had their own complexity and their own production rather than being simple dependencies. Which reframes the whole thing: not a few great cities with a hinterland, but a dense landscape of settlements of many sizes, with the cities as the largest nodes. And it fits the ending. A society whose weight is in its villages does not collapse when its cities empty. It redistributes. Next lesson: the bill for all of this.

Shot list

One AI shot, thirteen code-rendered shots, six museum or research images. This lesson leans hardest on real photography of any in the course, which is deliberate: the absences are only convincing if the viewer sees the sites.