Where We Left Off
This part covers around a hundred and fifty million people and eleven previous parts have barely mentioned them. Chapter Nine concedes the insurgency's own record outright, and Chapter Ten ends on the one gap in this series that nobody had to decide to create.
Every part of this series has found the same instruments used earlier and harder where the state’s presence was in question. This part is about the places where it was never established at all.
Five findings carry into this part.
The state ledger. Part Five committed every subsequent part to record which places its material affected, and its own ledger recorded the north-eastern states as absorbing refugee movements twice and receiving central attention chiefly as a security frontier. This part takes that entry properly.
Wealth is relational. Part Five’s central finding: a policy that removes one region’s advantage does not create wealth, it moves it, and the winner experiences success while the loser experiences failure and neither sees the pipe. Chapter Four is that finding applied to land.
An emergency power with no expiry becomes the ordinary law. Part Two found it in preventive detention, Part Three tracked it through a decade, Part Nine found it at full extension. Chapter Five is the most extreme case in the series: sixty-seven years.
When a state that counts everything has no count of something, somebody decided that. Part Two’s rule, now applied four times. Chapter Seven is the fifth and it produces a variant of the rule rather than another instance of it.
Instruments used earlier and harder where the state’s presence is in question. Part Seven found it in Kashmir, Part Ten in Punjab. Chapter Six asks whether it holds here, and the answer arrives as three maps.
One thing about this part’s existence. Twelve parts of this series have covered elections, amendments, wars, judgments and scandals, and have mentioned the people in this part almost not at all. That is not an oversight I am correcting; it is the ordinary shape of Indian political history, and it is itself one of the findings.
How to read this book
The Six Boxes
Six coloured boxes run through the series, each doing one job. Here they are with examples from the material ahead.
A Word Box explains a hard word the moment it first appears.
Scheduled Tribe: a legal category. Communities listed by presidential order under Article 342 of the constitution, entitled to reservation in legislatures, education and employment, and to specific protections over land.
The word “tribe” carries assumptions the category does not support. These are several hundred distinct communities, speaking different languages, with different histories, spread from the Nilgiris to the Himalaya, and their only shared feature is a legal designation and its consequences.
Many prefer Adivasi — “original inhabitant” — which is a political term rather than a legal one, and is contested precisely because of what it claims.
Why it matters here: around a hundred and fifty million people carry this designation, and almost everything in this part follows from what the designation was supposed to protect and did not.
An In Real Terms box turns a number too big to picture into something with a body.
The most cited estimate of people displaced by development projects in India since 1947 is around sixty million.
That is more people than live in Italy. It is roughly the population India had in the Bombay Presidency at independence. Moved, from land they lived on, for dams, mines, industrial plants, power stations, roads and protected forests.
Around a quarter were resettled in any organised way.
And the figure is not the Indian government’s. It is a scholar’s estimate, assembled project by project from state records. There is no official national count of the people the Indian state has displaced.
A How We Actually Know This box shows the physical evidence and then says what it cannot prove.
The evidence in this part is worse than in any other, and the reason is the subject.
Strong: the law. The Fifth and Sixth Schedules, the Panchayats (Extension to Scheduled Areas) Act 1996, the Forest Rights Act 2006, the Armed Forces (Special Powers) Act 1958 and its notifications, and the judgments in Samatha and Niyamgiri. All published, all checkable.
Strong: the census. Scheduled Tribe populations, district by district, decade by decade.
Weak or absent: almost every number about what was done. Displacement, land alienated, land restored, deaths in operations, compensation actually paid. These are estimates by researchers, or state-level figures that have never been added up.
Which produces the asymmetry that governs this whole part: India has excellent records of what it promised these people and almost none of what it did to them.
An Argument box appears where serious people disagree, each side at its strongest.
The framing question, and the case for the state is stronger than it is usually allowed to be.
Every industrialising country has moved people for dams, mines and cities, and none has done it gently. India in 1947 had famines within living memory, no power, no steel and no irrigation beyond the colonial canals. The dams and mines built since have fed and powered a country of a billion and a half. The alternative to displacing people from a valley is not a happier valley; it is no reservoir, no electricity, and a different set of people paying with their lives in a drought. That is a real trade and the people who denounce it have generally been on the receiving end of the benefits.
Nobody disputes that development imposes costs. The objection is to who bore them and how it was decided. Scheduled Tribes are around eight per cent of India’s population and something between forty and sixty per cent of those displaced. That is not a trade-off, it is a distribution, and it was produced by a rule: land held communally, without individual title, in areas the state classified as its own forest, is the cheapest land in India to take. The people who paid were selected by a property system, not by a calculation about the common good.
What would settle it: a project-by-project comparison of where the output went against where the displaced people ended up. The material exists in project files. Nobody has assembled it.
Why people care so much: because the first argument is the one every Indian government of every party has made, and the second is the argument of an insurgency that has cost tens of thousands of lives.
A Hidden Assumption box digs out a belief sitting underneath an argument that both sides accept without discussing.
Everybody who writes about these regions — sympathetically or otherwise — assumes that they are remote.
It is in the vocabulary. Remote areas. Interior districts. Backward regions. The last mile. Places the state has not yet reached, where development has not yet arrived, where administration is thin because the terrain is difficult and the population is scattered.
Look at what is actually in them.
India’s coal is there. Its iron ore, its bauxite, its manganese and its uranium are there. The rivers that were dammed run through them. Roughly a fifth of India’s forest cover, and the catchments that supply the reservoirs the plains depend on.
A place from which a country extracts its coal, its iron and its electricity is not remote. It is central, and the state’s presence in it is not thin — it is enormous, and it consists of mines, dams, forest departments, project authorities and armed police.
The word “remote” describes one specific thing accurately: the absence of the parts of the state that deliver rather than take. Schools, clinics, courts, land records, banks. Those are thin. The extractive apparatus is not.
So the standard prescription — that these places need development to reach them — misdescribes the situation completely. Development reached them first. It reached them in the 1950s, before it reached most of India, because that is where the coal was.
What did not reach them was the other half of a state: the half that records who owns what, hears a complaint, enforces a protection and counts the dead. Every chapter of this part is about the gap between those two halves, and the gap is not a matter of distance.
And every chapter closes with a Remember This box: the chapter in the plainest words available.
Around sixty million people displaced since 1947, about a quarter resettled — and the figure is a scholar’s estimate, not the government’s. There is no official national count.
India has excellent records of what it promised these people and almost none of what it did to them.
These places are not remote. The coal, iron, bauxite and dams are there. What is thin is not the state but the half of it that delivers rather than takes.
1Who This Part Is About
More people than live in Russia. Twelve parts of this series have mentioned them roughly twice, and that is the ordinary proportion.
1.1 — The numbers
Around eight and a half per cent of India’s population is classified as Scheduled Tribe. On present figures that is something in the order of a hundred and twenty million people, and the count is conservative because the classification is a legal one and not everybody who lives as these communities live is on the list.
Add the populations of the north-eastern states, where the Sixth Schedule and separate constitutional arrangements apply, and the total this part concerns is around a hundred and fifty million.
Where they are, in outline:
Central and eastern India. A belt running roughly from Gujarat and Rajasthan through Madhya Pradesh, Maharashtra and Chhattisgarh into Jharkhand, Odisha, Telangana and Andhra Pradesh. This is where the Fifth Schedule applies, in ten states.
The north-east. Assam, Meghalaya, Tripura and Mizoram under the Sixth Schedule; Nagaland, Arunachal Pradesh and Manipur under separate provisions and separate histories.
Elsewhere — the Nilgiris, the Andaman and Nicobar Islands, pockets across the Himalaya and the Western Ghats.
The population figures in this chapter are the strongest numbers in the whole part, and it is worth being clear why they are so much better than everything that follows.
The census. Scheduled Tribe population has been enumerated at every decennial census, by district, with the classification fixed by presidential order. It is one of the oldest continuous statistical series in India.
Which produces the asymmetry this part keeps returning to. The Indian state knows precisely how many of these people there are and where they live, to the village. It has known for a century.
What it does not know is what has happened to them, and Chapter Seven is the list.
What the census cannot show: the boundaries of the category. Inclusion in the list is a political and administrative decision, revised by order, and communities have been added and have campaigned for decades to be added. So the figure is exact about a legal class and approximate about a social reality — which is why this chapter gives a range once the north-east is included.
Solid: the population, the share, the district distribution.
1.2 — Why they have not appeared
Twelve parts of this series have covered elections, constitutional amendments, wars, judgments, massacres and scandals, and have barely mentioned any of this.
That is not carelessness and it is worth being exact about the mechanism, because the mechanism is the subject.
Electoral arithmetic. A hundred and twenty million people spread across ten states in scattered districts do not decide national elections. Seats are reserved, which guarantees representation and does not guarantee leverage: a reserved seat is still won by a party that needs a national majority, and its holder is bound by Part Twelve’s Tenth Schedule.
Absence from the record. Almost every part of this series has been built on documents — judgments, commission reports, gazettes, parliamentary answers. Those documents exist where the state’s recording apparatus operates. The front matter’s hidden assumption is the reason it operates thinly here.
No single event. Indian political history is organised around events with dates. What this part describes has no date. It is a continuous condition, and continuous conditions do not enter a chronology.
Which is the honest reason this is Part Thirteen and not Part Two.
Scheduled Area: territory formally designated under the Fifth Schedule, where special rules about land apply.
The designation is not the same as where Scheduled Tribes live. Large tribal populations live outside Scheduled Areas and receive none of the land protections; and the boundaries, drawn from colonial classifications, have been revised rarely.
That gap does quiet work throughout this part. A protection attached to a place rather than to a person protects whoever is standing in the right square, and the squares were drawn in 1950 on nineteenth-century lines.
Why it matters here: it is one of the reasons the figures in Chapter Seven would be hard to compile even if somebody wanted to — the protected category and the affected population are not the same set.
The scale is the part most readers have never had put to them plainly.
A hundred and fifty million people is more than the population of Russia, or Japan, or Mexico.
If it were a country it would be the ninth or tenth largest on earth.
Now consider that this series has produced twelve books on Indian public life since 1947, running to well over two hundred thousand words, and that before this page they have contained perhaps two dozen sentences about these people.
That proportion is not unusual. It is close to the proportion in most general histories of independent India, and in the syllabus of most Indian schools.
Which raises a fair objection to this part’s own framing, and it should be dealt with before the chapters rather than after.
This part puts the Nagas, the Gonds, the Santhals, the Bhils and the Dongria Kondh in one book. That is a choice and it is criticised.
A Naga village in the hills and a Bhil settlement in Rajasthan share a legal designation and almost nothing else — not language, not religion, not economy, not history, not what happened to them. The north-east’s grievance is about political status and identity; central India’s is about land and extraction. Treating them together produces exactly the abstraction that has made these communities invisible: a category rather than peoples, which is how administrations have described them since 1874 and part of why nothing specific ever gets done.
The unity in this part is not cultural, and it is not claimed to be. It is that the Indian state has applied a common set of instruments to all of them — a common constitutional category, a common property regime that recognises only recorded title, a common pattern of acquisition, and in several regions a common special powers Act. When one actor uses the same tools on many different people, the tools are a legitimate subject, and describing them requires putting the cases side by side.
What would settle it: nothing evidential. It is a question about how to write, not about what happened.
Why people care so much: because the flattening is real and has consequences, and because the alternative — several hundred separate accounts — is how a subject stays permanently unaddressed.
With that settled, the rest of this part proceeds on the second position, and the reader should hold the first as a running caution.
Around 8.5 per cent of India’s population is classified as Scheduled Tribe — roughly 120 million people, and about 150 million including the north-eastern populations this part covers.
They are in a belt from Gujarat to Odisha under the Fifth Schedule in ten states, and in the north-east under the Sixth Schedule and separate arrangements.
Three reasons they have not appeared: they do not decide national elections, the documentary record is thin where the recording apparatus is thin, and there is no single dated event. Continuous conditions do not enter a chronology.
2What Was Promised
The constitution contains two entire Schedules written to protect these people’s land. They are not weak documents. What they are is a set of protections handed to the wrong people to enforce.
2.1 — The Fifth Schedule
The Fifth Schedule applies to Scheduled Areas in ten states of central and eastern India. Its central provisions are two.
The Governor may make regulations prohibiting or restricting the transfer of land by members of Scheduled Tribes, regulating moneylending to them, and regulating the allotment of land.
A Tribes Advisory Council is to be constituted in each state, consisting largely of members of the legislature belonging to Scheduled Tribes, to advise on their welfare and advancement.
The Governor is also required to report annually to the President on the administration of the Scheduled Areas, and the Union may give directions to the state about it.
Read as text, that is substantial. A constitutional power to prohibit land transfer, held by a central appointee rather than by the state government, with a reporting duty attached.
2.2 — The Sixth Schedule
The Sixth Schedule applies to tribal areas in Assam, Meghalaya, Tripura and Mizoram, and it is a fundamentally different instrument.
It creates Autonomous District Councils — elected bodies with actual legislative power over land, forests other than reserved forests, water for agriculture, inheritance, marriage, social custom, village administration and the appointment of village-level authorities. They can levy certain taxes and constitute village courts.
That is not an advisory body. It is a subordinate legislature with a defined field, and it is the most substantial devolution anywhere in the Indian constitution.
Chapter Eight returns to what these councils have actually delivered, because the answer is more positive than anything else in this part.
2.3 — What was added later
Two statutes matter. One was passed by a Congress-led government and one was not, and this part says which.
The Panchayats (Extension to Scheduled Areas) Act, 1996, generally called PESA, enacted in December 1996 under a United Front government. It extended the panchayat system of Part Twelve’s Chapter Seven to Fifth Schedule areas with modifications, and it gave the gram sabha — the assembly of every adult in a village — powers that are, on paper, among the strongest held by any local body in India.
The gram sabha is to be consulted before land acquisition. It has the power to prevent alienation of tribal land and to take action to restore land unlawfully alienated. It owns minor forest produce. It is to be consulted before minor mineral leases.
The Forest Rights Act, 2006, enacted under a Congress-led government. It recognises rights that were never recorded: individual occupation rights over forest land, and community rights over forest resources, including ownership of minor forest produce. It requires gram sabha consent before forest land is diverted.
The Act’s preamble contains an admission unusual in Indian legislation: that these communities’ rights were not recorded during the consolidation of state forests, and that a historical injustice resulted.
2.4 — Who enforces it
And here is the whole of this chapter.
The Fifth Schedule’s land protection is exercised by the Governor, who acts on the advice of the state government.
PESA required state governments to make rules bringing their panchayat laws into conformity. Many took a decade or more. Several diluted the provisions in the rules.
Restoration of alienated land is the responsibility of the state revenue administration.
Diversion of forest land requires the consent of a gram sabha, certified by the district administration.
In every case, the body that must enforce the protection is the body that wants the land — because the state government is the licensor of the mine, the promoter of the dam, and the recipient of the royalty.
Gram sabha: the assembly of all adult residents of a village, as distinct from the elected panchayat, which is a committee.
The distinction matters enormously in this part. A panchayat can be captured — five or seven people can be persuaded, bought or pressured. A gram sabha is everybody, meeting in one place, and it is the hardest body in Indian public life to buy.
Which is why PESA and the Forest Rights Act vest their consent powers in the gram sabha and not the panchayat, and why Chapter Eight’s one clear success came from twelve of them voting.
Why it matters here: it is also why the fight over these laws has been about who certifies that a gram sabha met and what it decided.
The difference between the two Schedules is easiest to see if you ask what each one actually puts in a person’s hands.
Compare the two Schedules as instruments a person could actually use.
Under the Sixth: your district council makes the law on land. You elect it. If it allows something you object to, you vote it out at the next election.
Under the Fifth: the Governor may make a regulation. The Governor acts on the advice of the state government. You do not elect the Governor. The Tribes Advisory Council advises and cannot decide. If a regulation is not made, or is not enforced, there is nobody you can remove.
Both are in the same constitution, written by the same assembly, in the same year, for the same purpose.
One of them gives the protected population an instrument it can operate. Chapter Eight records which one still has its land.
Which raises the obvious question about why the two were written so differently.
The comparison is unavoidable once the two are set side by side, and there is a real case on both sides.
The Sixth Schedule areas were compact, remote, overwhelmingly tribal, and had been administered separately under colonial rule as excluded areas. Autonomous councils were workable there because the territory was homogeneous. The Fifth Schedule areas are scattered pockets inside large states with mixed populations, often a minority even within their own districts. Carving out subordinate legislatures across seven states with mixed populations would have created a permanent boundary dispute in each, and the constituent assembly judged that a protective power exercised from above was more practical than a devolved one.
The Sixth Schedule areas had recently been in armed or political conflict over their status, and had negotiating strength; the Fifth Schedule populations did not. That is a better predictor of which arrangement each got than any assessment of workability. And the practical objection is answered by what actually happened: seventy-five years later, one arrangement has kept its land and the other has not, which is a heavy price for administrative convenience.
What would settle it: the constituent assembly debates on the two Schedules, which are published and have never been analysed comparatively on this question.
Why people care so much: because proposals to extend Sixth Schedule status to areas in central India are live right now and are resisted on exactly the first argument.
Underneath the whole discussion, though, is a way of describing what went wrong that is shared by people who agree on nothing else.
Everybody who writes about tribal land in India assumes that the Schedules failed to protect.
Critics say the protections were inadequate, unimplemented, a dead letter. Defenders say the law is strong and enforcement is imperfect, as it is with all Indian law. Both are arguing about how well a protection worked.
The protections did not fail. They were never given to anybody with an interest in enforcing them.
Go through section 2.4 again as a question about incentives rather than about capacity. Who may prohibit land transfer? A Governor, who acts on state advice. Who makes the rules under PESA? The state. Who restores alienated land? The state’s revenue officers. Who certifies that a gram sabha consented? The district administration.
And who grants the mining lease, promotes the irrigation project, receives the royalty and answers to an electorate in which these communities are a small minority? The same state government.
That is not weak enforcement. It is an enforcement architecture in which the protected party has no instrument it can operate itself, and the party with the contrary interest holds every one.
Which explains a pattern that otherwise looks like coincidence: every genuine advance in this area has come from a body outside that architecture. The Supreme Court in Samatha. The Supreme Court in Niyamgiri. And a gram sabha, which is the one decision-making body in the chain whose membership is the protected population itself.
The general rule this series keeps arriving at from different directions: a right is worth the independence of whoever enforces it. Part Eleven found it in a police force investigating itself. This is the same finding in property law, and it is seventy-five years old.
Which is the chapter, and everything that follows in Part Thirteen is what happened underneath it.
The Fifth Schedule lets a Governor prohibit transfer of tribal land in ten states. The Sixth Schedule creates Autonomous District Councils with real legislative power over land — the most substantial devolution in the constitution.
PESA 1996 gave gram sabhas power to prevent alienation and restore land. The Forest Rights Act 2006 admits in its preamble that these rights were never recorded and that a historical injustice resulted.
The protections did not fail. They were handed to the party with the contrary interest. The state that must enforce them is the state that licenses the mine and takes the royalty.
3The Land That Moved Anyway
Five mechanisms, none of them illegal, by which land the constitution protects has changed hands for seventy-five years.
3.1 — The mechanisms
Land does not usually leave these communities by a transaction anybody would call theft. It leaves by procedure, and the procedures have names.
Eminent domain. Until 2013 the Land Acquisition Act of 1894 — a colonial statute — permitted the state to take any land for a public purpose on payment of compensation assessed at market value. Where there is no market because land is held communally and rarely sold, the assessed value is whatever the collector determines. The Act required no consent from anybody.
Forest classification. Large areas that communities had occupied for generations were recorded as state forest during and after colonial consolidation, without their rights being entered. The Forest Rights Act’s preamble says this. The consequence is that people living on land for a century were, on the record, encroachers on government property.
Benami and the fraudulent transfer. Where transfer to a non-tribal is prohibited, the transaction is conducted through a tribal name-holder, or as a long lease, or as a mortgage that is never redeemed. Restoration requires the revenue administration of section 2.4 to act.
Debt. Land passing to a moneylender against a loan, at a valuation set by the lender. The Fifth Schedule expressly empowers the Governor to regulate moneylending, which is a measure of how old and well understood the problem is.
Protected areas. National parks and sanctuaries declared over inhabited land, with residents relocated as a condition of conservation.
Note what these have in common. Not one requires anybody to break a law. Each is a lawful procedure operating on a population without recorded title, in a system where title is the only thing a procedure recognises.
Consider what a land record actually does.
If your family has farmed four acres for a hundred and fifty years and there is a paper saying so, then acquisition means a notice, a valuation, a hearing and a cheque. The paper does not stop the acquisition. What it does is make you visible to the procedure.
If there is no paper, you are not a landowner losing land. You are, on the file, a person occupying government forest.
Same field. Same century and a half. The entire difference between compensation and eviction is one entry in a register that somebody was supposed to make in about 1880 and did not.
Almost everything in this part follows from that one absence, which is why the 2006 Act’s preamble had to begin by admitting it.
3.2 — Samatha, and after
In 1997 the Supreme Court decided Samatha v. State of Andhra Pradesh.
The question was whether a state government could grant mining leases over land in a Scheduled Area to non-tribal private companies. The state’s argument was that the prohibition on transfer applied to persons, and that the government was not a person.
The Court held that it was. Land in Scheduled Areas, including government land, could not be leased to non-tribals for mining. Where minerals were to be extracted, it should be by the communities themselves or by cooperatives, with state support.
It is one of the strongest judgments on tribal land anywhere in Indian law.
What followed is instructive. Its application was confined largely to Andhra Pradesh. Other states took the position that it did not bind them. Attempts were made to amend the Fifth Schedule to overcome it. Mining leases in Scheduled Areas across the belt continued to be granted.
A judgment that requires an unwilling state government to act against its own revenue interest is a judgment that requires somebody to bring a fresh case in every state. Chapter Nine returns to this.
Alienation of land: the legal term for land passing out of the hands of a protected group, whether by sale, lease, mortgage, fraud, encroachment or acquisition.
It is a deliberately wide word and the width is the point. Indian land law needed a term covering every route by which the thing happened, because the routes are many and only some of them involve anybody doing something unlawful.
Note what the word does not say. It does not say who took the land, or whether they were entitled to, or whether anything can be done about it. It describes a movement and assigns no responsibility, which is convenient for a statute and much less convenient for a person trying to find out what happened to a particular field.
Why it matters here: “land alienated” is the heading under which the missing figure in section 3.3 would sit, if anybody had ever compiled it.
3.3 — What the numbers are
Here this chapter stops, and the stopping is the point.
There is no national figure for how much land protected by the Fifth Schedule has been alienated. There is no national figure for how much has been restored.
State-level figures exist, in revenue records, in occasional committee reports and in scattered studies. They have never been compiled.
Chapter Seven is about that sentence, which appears in this part more often than any other.
The mechanisms in section 3.1 are documented in an unusual way: by the laws written to stop them.
You do not empower a Governor to regulate moneylending unless land is being lost to moneylenders. You do not give a gram sabha power to restore unlawfully alienated land unless land is being unlawfully alienated. You do not legislate in 2006 that rights were never recorded unless they were not recorded.
Every mechanism in this chapter has a statutory provision aimed at it, which means the Indian state has identified each one, in writing, and said so.
Solid: Samatha and its terms; the text of the 1894 Act and its absence of a consent requirement; the Forest Rights Act preamble.
What none of it can show: quantity. This chapter can establish what happened and cannot establish how much, and section 3.3 says so rather than reaching for an estimate.
There is also a serious argument that the protections themselves are part of the problem, and it should be heard at full strength.
A serious argument holds that the protections are themselves the trap. It deserves stating properly because it is made by people who want these communities better off.
Land that cannot be sold cannot be mortgaged, and land that cannot be mortgaged cannot raise capital. The prohibition on transfer freezes a community’s principal asset out of the economy, which means the owner of forty acres in a Scheduled Area cannot borrow against it to educate a child or start a business, while an owner of two acres in Punjab can. The protection has kept the land and kept the poverty, and the market has found its way round it anyway through the devices in section 3.1 — at prices far below what an open sale would fetch.
The prohibition exists because the alternative was tried. Wherever these restrictions were absent or lapsed, land moved rapidly and did not come back, and the sellers did not become capitalists — they became labourers on land they had owned. The problem with the argument from capital is that it assumes the seller can convert a lump sum into a livelihood, and the record of people who received compensation for acquired land is the direct evidence against it: a quarter were resettled and the money did not substitute for the land.
What would settle it: comparing outcomes where forest rights titles have actually been issued against comparable areas where they have not. The data is being generated now and nobody has analysed it.
Why people care so much: because the first argument is used to justify removing the protections and the second to justify keeping a system that has not worked, and both sides use the other’s failure as their own case.
What the chapter establishes, and what it declines to, comes to this.
Five mechanisms, none of them illegal: eminent domain under an 1894 statute requiring no consent; forest classification that recorded occupants as encroachers; benami transfers; debt; and protected areas.
Samatha (1997) held that government land in Scheduled Areas could not be leased to non-tribals for mining. Its application was confined largely to one state and leases continued elsewhere.
There is no national figure for how much protected land has been alienated, or how much restored. That sentence appears in this part more than any other.
4Displacement
Sixty million people. The figure is the most cited in the field, it is probably conservative, and it does not come from the Indian government.
4.1 — Where the number comes from
The estimate generally used for people displaced by development projects in India between 1947 and 2000 is more than sixty million. It comes from the work of Walter Fernandes, a researcher who spent decades compiling state-level databases.
An earlier and narrower estimate by the same researcher with a colleague put the figure at at least 21.3 million for a shorter period and a narrower set of project types. The gap between the two is not a disagreement; it is the difference between what had been compiled at two points in time, and it indicates the direction the number moves as more records are examined.
Be clear about what these figures are. They were assembled project by project and state by state, from land acquisition records, project reports and district files, by researchers who went and looked.
They are not a government count. There is no government count. Chapter Seven is about that.
4.2 — What it rests on and what it does not
The estimate’s strength is that it is built from the state’s own project records, which means each component is documented even though the total is not.
Its weaknesses are three and this part states them rather than glossing them.
Definitions vary. Displaced persons — those physically moved — and project-affected persons, who lose land or livelihood without moving, are counted differently in different studies.
Coverage is uneven. Some states have been studied thoroughly and others barely.
Official records understate. Land acquisition files record title-holders. A household with no recorded title — which, as Chapter Three established, describes a very large share of the people concerned — may not appear in the file at all.
That third point matters most, and it runs one way. It means sixty million is a floor.
The sixty million figure deserves a full account of its provenance, because this part leans on it and because it is the most attacked number in the field.
Method: state-by-state databases built from land acquisition awards, project completion reports, district gazetteers and departmental records — that is, from the state’s own paperwork, examined by researchers who went to the offices.
Not a projection. It is not a rate applied to a population or a model extrapolated from a sample. It is a sum of counted components, which is why it rises as more states are covered rather than moving in both directions.
The honest weaknesses are in section 4.2 and this part states them rather than defending the number.
What it cannot show: outcomes. The databases record how many were displaced, not what happened to them afterwards. Everything in section 4.4 comes from separate studies of particular projects, which are consistent with each other and are not a national sample.
And the reason a researcher had to do this at all is Chapter Seven.
4.3 — Who they were
Between 1951 and 1990, on these estimates, more than 8.5 million Scheduled Tribe people were displaced.
Set that against the population share. Scheduled Tribes are around eight and a half per cent of India, and they are commonly estimated at something between forty and sixty per cent of those displaced.
That is a ratio of roughly five or six to one.
The cause is not mysterious and it is not malice. It is the intersection of two facts: the minerals and the dam sites are where these communities live, and their land is the cheapest in India to acquire, because Chapter Three’s mechanisms operate on land without recorded individual title.
4.4 — What resettlement means
Roughly a quarter of those displaced were resettled in any organised way. The remaining three-quarters were not.
And “resettled” needs unpacking, because the word conceals the whole of the problem.
A cultivator displaced from land receives compensation assessed on the land. He does not receive land. Whether he can buy comparable land with the money depends on whether comparable land is available near where he now is, which after a reservoir has filled it generally is not.
A family whose income came from forest produce — which is not owned, not titled and not valued — receives nothing for it, because it was not on the file.
A community moved is a community dispersed, and the arrangements that made subsistence possible — grazing rights, shared labour, the network of relations Chapter Three describes — do not travel.
Which is why the outcome that recurs in every study is the same: a landholder becomes a labourer. Not immediately, and not because the compensation was zero, but because a lump sum is not a livelihood and the thing that was taken was a livelihood.
Rehabilitation, resettlement and compensation: three different things, routinely used as if they were one.
Compensation is money for what was taken, valued by the acquirer.
Resettlement is being given somewhere to live.
Rehabilitation is being restored to a comparable livelihood — the only one of the three that measures whether the person is as well off as before.
Why it matters here: the figure of roughly a quarter is for resettlement. The proportion rehabilitated — restored to a comparable living — has never been established nationally for any period, and the project-level studies that exist put it far lower.
The scale is easier to hold against something familiar.
Sixty million over fifty-three years is more than a million people a year, every year, for half a century.
It is roughly one in every twenty Indians alive today, or their parents or grandparents.
Set beside the largest displacements in modern history it sits in the same category as Partition — which moved somewhere between ten and twenty million people over a few years and is the defining trauma of the Indian twentieth century, taught in every school.
This is three to six times larger, took place over five decades, and has no name.
All of which rests on a way of describing the thing that almost nobody examines, including the people who denounce it.
Everybody who argues about this — those who defend the dams and those who oppose them — assumes that displacement is a cost of development.
The defenders say it is a cost worth paying. The critics say the cost was too high or was paid by the wrong people. Both are using the vocabulary of a price: something was bought, and this is what it cost.
A cost is borne by the person who obtains the benefit. That is what makes it a cost rather than something else.
Part Five established the vocabulary for what this actually is. It is a transfer. The reservoir irrigates a plain a hundred and fifty kilometres away. The coal fires a power station in another state. The bauxite becomes aluminium in a smelter somebody else owns. The output goes to people who are not the people who moved.
And the check is the same one Part Five used. If it were a cost, the districts that bore it would have grown richer as the projects matured, because the benefit would have accrued where the price was paid. Chapter Six shows what actually happened to those districts.
The framing does real work for both sides, which is why neither abandons it. For the defenders, “cost” implies necessity and a calculation somebody made responsibly. For the critics, “cost” allows the remedy to be compensation — pay a fair price and the transaction becomes just.
But if it is a transfer, compensation is not the remedy, because the question is not whether the price was fair. It is why one population supplies the inputs and another receives the output, and no payment answers that.
Part Five’s finding, arriving in a new form: the winner experiences success, the loser experiences failure, and neither of them sees the pipe.
Chapter Six puts that claim to a test with dates and district figures. First, the instrument used where the transfer was resisted.
More than sixty million displaced between 1947 and 2000. The figure is a researcher’s, assembled project by project from state records. It is a floor, because households without recorded title do not appear in acquisition files.
Scheduled Tribes are about 8.5 per cent of India and something between forty and sixty per cent of the displaced. About a quarter were resettled in any organised way.
This is not a cost, it is a transfer. A cost is borne by whoever gets the benefit, and the output went somewhere else.
5Sixty-Seven Years
It descends from an ordinance the British made in 1942 to crush the Quit India movement. It has been in force somewhere in India continuously since 1958.
5.1 — Where it comes from
On 15 August 1942 the colonial government promulgated the Armed Forces Special Powers Ordinance, to suppress the Quit India movement.
Five years later that framework was invoked for the disturbances of Partition. In 1955 Assam enacted a Disturbed Areas Act. In May 1958 the President promulgated the Armed Forces (Assam and Manipur) Special Powers Ordinance, and on 11 September 1958 Parliament replaced it with an Act.
The date deserves noting for what else was happening. Part Three’s Chapter Two describes the same Parliament, in the same years, using its majority to amend the constitution repeatedly. Part Two’s Chapter Nine describes preventive detention being renewed. This was a legislature entirely comfortable with special powers.
In 1972 the Act was amended so that the central government could declare any area disturbed, without the consent of the state concerned.
Hold that amendment. It is the reason section 5.4 ends as it does.
5.2 — What it permits
In an area declared disturbed, any commissioned or non-commissioned officer may:
Fire upon or otherwise use force, even to the causing of death, against a person contravening a prohibitory order or carrying weapons, after giving such warning as he considers necessary.
Arrest without warrant, on suspicion, using such force as may be necessary.
Enter and search any premises without a warrant.
Destroy structures used as hideouts or as arms dumps.
And then section 6, which is the operative provision of the entire Act:
No prosecution or other legal proceeding may be instituted against any person for anything done under the Act except with the previous sanction of the central government.
5.3 — Where, and for how long
It has been withdrawn from Mizoram in the 1980s, from Tripura in 2015, and from Meghalaya in 2018. Those are real and they belong in Chapter Eight.
As of the most recent notifications, it remains in force in parts of Nagaland, Manipur, Arunachal Pradesh and Assam, extended by six-month declarations — the latest running to March 2026. It applies separately in Jammu and Kashmir under a 1990 Act.
Which means: an Act described as extraordinary has been continuously in force somewhere in India for sixty-seven years, renewed every six months, in territory whose inhabitants are Indian citizens with the same constitutional rights as everybody else.
Sixty-seven years is long enough for a child born under the Act in Nagaland in 1958 to have become a grandparent under it.
It is longer than the Emergency of Part Nine by a factor of about thirty-two. Longer than every war India has fought, added together, many times over.
It is longer than the entire period from the Act’s ancestor in 1942 back to the First World War.
And it is renewed by a notification of a few lines, every six months, one hundred and thirty-odd times, without a debate, a report or a reason.
5.4 — What has been established, and what happened to it
Four bodies and a court have examined it.
In 1997, in Naga People’s Movement of Human Rights, the Supreme Court upheld the Act’s constitutional validity while laying down safeguards on the use of force and the investigation of complaints.
In 2005 a committee under Justice B.P. Jeevan Reddy, appointed by the government, recommended repeal. It described the Act as too sketchy, too bald and quite inadequate, and proposed that necessary provisions be inserted into ordinary anti-terror legislation instead. The report was not acted on.
From 2000 to 2016, Irom Sharmila conducted a hunger strike in Manipur demanding repeal. Sixteen years.
In 2013 a committee under Justice Santosh Hegde, appointed by the Supreme Court, examined a sample of alleged encounter killings in Manipur and found that none of those it examined was a genuine encounter.
In 2016, in the case brought by families of those killed, the Supreme Court held that there is no blanket immunity, that the armed forces cannot claim absolute protection, and that every death caused by them must be investigated where there is a complaint of excess.
And then December 2021, in Mon district, Nagaland, where security forces killed thirteen civilians and a fourteenth died the following day. The state cabinet formally recommended repeal. The Chief Minister called it a draconian law and pointed out that the insurgent groups had been in ceasefire for around twenty-five years.
The Nagaland government sought to prosecute the personnel involved. Sanction was refused under section 6.
And the six-month declarations continued.
Sanction to prosecute: a requirement that a designated authority give permission before a public servant can be tried for something done in the course of duty.
It is not unusual in itself. Most legal systems protect officials from vexatious prosecution for official acts, and Indian criminal procedure contains a general version of it.
What section 6 does differently is two things. The sanctioning authority is the central government — not a court, not the state whose citizens are affected. And there is no time limit, no duty to give reasons, and no effective appeal against refusal.
Why it matters here: a right without a remedy is not a right, and this is the provision that converts everything in section 5.2 from a regulated power into an unreviewable one. It is also why the 2016 judgment mattered and why Mon shows its limits.
| Year | Body | What it found | What followed |
|---|---|---|---|
| 1997 | Supreme Court | The Act is constitutionally valid, subject to safeguards on the use of force and the investigation of complaints. | Declarations continued. |
| 2005 | Jeevan Reddy Committee | Repeal. The Act is too sketchy and inadequate; move what is needed into ordinary anti-terror law. | Not acted on. |
| 2013 | Santosh Hegde Committee | None of the sampled Manipur encounters was genuine. | Declarations continued. |
| 2016 | Supreme Court | No blanket immunity. Every death must be investigated where excess is alleged. | Declarations continued. |
| 2021–23 | Nagaland state cabinet | Recommended repeal after fourteen civilian deaths at Mon; sought to prosecute. | Sanction refused under section 6. |
Which is a record with an unusual property: nothing in the right-hand column varies.
The professional case is not a bad-faith case and it is rarely stated properly by people who want repeal.
A soldier is not a policeman. He is trained to close with and kill an enemy, not to make arrests, preserve a scene and give evidence. If you send the army into an internal conflict — which a government does only when the police have already failed — you must either give it a legal framework matched to what it is trained to do, or accept that every operation exposes a nineteen-year-old to prosecution for following an order. Section 6 exists because officers will not commit troops where the legal position is that each of them is personally liable afterwards. Remove it and the practical result is not gentler operations; it is that the army refuses the task and the conflict is handled worse.
The argument proves at most that a framework is needed, not that this one is. The government’s own committee said exactly that in 2005 — repeal the Act and place what is genuinely required in ordinary law, with safeguards. And the practical claim is testable: the Act has been withdrawn from three states and the armed forces did not refuse to operate in any of them. What the professional case cannot reach is the actual complaint, which is not that immunity exists but that it is granted by the executive, without reasons, without a time limit, and against a state government’s formal request.
What would settle it: what actually happened to operational effectiveness in Tripura and Meghalaya after withdrawal. Nobody has published it.
Why people care so much: because one side is arguing about a nineteen-year-old in a patrol and the other about a family at Mon, and both of those are real.
Which leaves a question about what kind of law this actually is.
Everybody who argues about this Act — those demanding repeal and those defending it — assumes that it is an emergency power.
Its defenders say extraordinary situations require extraordinary measures. Its critics say the emergency has been manufactured or has passed. Both accept that the Act belongs to a category of law that is exceptional, temporary in principle, and to be judged by whether the exceptional conditions obtain.
Sixty-seven years is not an exception. For everybody living in those districts it is simply the law, and has been for their entire lives, their parents’ entire lives, and in most cases their grandparents’ too.
The word doing the work is “disturbed” — a declaration renewed every six months, by a central government, on no stated criteria, subject to no external test, and in at least one case over the formal objection of the state cabinet.
Part Seven found precisely this operation performed by a different word. Article 370 was headed “temporary”, and Part Seven’s finding was that it was emptied over four decades while the heading stayed. Here the word is “disturbed”, and it has stayed for sixty-seven years while the conditions it describes have come and gone — insurgencies begun, negotiated, ended, and ceasefires held for a quarter of a century — without the declaration lapsing once.
This changes what the argument should be about. Asking whether the emergency justifies the powers assumes there is a mechanism by which the answer could matter. The 1972 amendment removed the state’s consent; section 6 removed the courts’ reach; four inquiries and a Supreme Court judgment have not altered a single declaration.
So the honest description is not an emergency power that has been overused. It is a permanent legal regime applying to particular territories and the people in them, renewed by a formality, which uses the vocabulary of emergency because that vocabulary is what makes it discussable.
Part Two’s rule about instruments with no expiry, at the furthest extension this series has found.
Sixty-seven years of one instrument, in short, comes to this.
It descends from a 1942 colonial ordinance against Quit India. The Act dates from 11 September 1958. The 1972 amendment let the centre declare any area disturbed without the state’s consent.
Section 6 bars any prosecution without the central government’s sanction. Repeal was recommended by a government committee in 2005 and the report was not acted on. The Supreme Court held in 2016 that there is no blanket immunity.
After Mon in December 2021 the state cabinet asked for repeal and sought to prosecute. Sanction was refused. Sixty-seven years is not an emergency; it is the law of those places.
6The Three Maps
Draw the Scheduled Areas. Draw India’s mineral belt. Draw where the insurgency has been strongest. They are almost the same map, and the order in which they were drawn settles what that means.
6.1 — The overlap
Take the ten states with Fifth Schedule areas and look at the districts.
The first map is the Scheduled Areas: a belt through southern Madhya Pradesh, Chhattisgarh, Jharkhand, northern and western Odisha, eastern Maharashtra, and parts of Telangana and Andhra Pradesh.
The second map is minerals. The bulk of India’s coal is in Jharkhand, Odisha and Chhattisgarh. So is most of its iron ore and its bauxite. The great river valley projects — Damodar, Mahanadi, Godavari, Narmada — run through or above these districts.
The third map is the insurgency. The districts described for two decades as the Red Corridor, where the Maoist movement has been strongest and where the security response has been heaviest.
The three maps overlap almost exactly. This is not disputed by anybody, and it is stated in Indian government documents.
6.2 — The order they were drawn in
Which of these came first decides everything about what the overlap means, so here is the sequence.
The minerals were always there.
The areas were scheduled in 1950, on the basis of colonial-era classifications going back to the nineteenth century. The designation recorded who lived there and that they required protection.
Extraction began immediately. The first great public sector projects of the 1950s — the steel plants, the Damodar Valley scheme, Hirakud, Rourkela — were built in these districts, in the first decade after independence. Part Four’s whole economic model was built on them.
The insurgency arrived last. The movement dates from 1967 in Bengal and became a significant presence in this belt from the late 1970s and through the 1980s — twenty to thirty years after the extraction started.
Royalty: the payment a mining company makes to the state for the mineral it extracts.
It goes to the state government, into the general revenues, and is spent wherever the state decides. It does not go to the district, the village or the people displaced, and until recently no mechanism required any share to return to the place the mineral came from.
A district-level fund financed from mining contributions was created in 2015, which is a real change and is one of the reasons this part does not say nothing was ever done.
Why it matters here: royalty is the reason section 2.4 ends where it does. The state government that must enforce the land protection is the state government whose budget depends on the lease being granted.
6.3 — What the districts look like now
Here is the test the front matter’s argument box named.
If displacement were a cost, borne locally and repaid in local development, then seventy years of supplying India’s coal, iron and electricity should have made these districts prosperous.
They are among the poorest districts in India. On almost every measure — income, literacy, infant mortality, access to electricity — the mineral districts of Jharkhand, Odisha and Chhattisgarh rank at or near the bottom, and have done for decades.
A district that has exported coal for seventy years and has villages without a connection is not a district that received a cost and a benefit. It is a district at one end of a pipe.
6.4 — The north-eastern version
The same test, applied differently.
In the north-east the instrument of Chapter Five did not follow the insurgencies; in the Naga case it arrived with one. So the sequence argument in section 6.2 does not transfer directly and this part will not pretend it does.
What transfers is the second half. The instrument has outlasted the conditions. Mizoram’s insurgency ended by accord in 1986 — Part Twelve’s Chapter Seven — and the Act had already gone. Nagaland’s principal groups have been in ceasefire since 1997, and the declarations continue into 2026.
Twenty-nine years of ceasefire, and a state cabinet asking for repeal, has not altered a six-month notification.
6.5 — The ledger for this part
| Where | What this part records |
|---|---|
| Jharkhand, Odisha, Chhattisgarh | The mineral belt. Fifth Schedule areas, the bulk of India’s coal, iron ore and bauxite, the largest share of displacement, and the districts where the insurgency has been strongest. Among the poorest in India after seventy years of extraction. |
| Madhya Pradesh, Maharashtra, Telangana, Andhra Pradesh | Fifth Schedule areas. Samatha arose in Andhra Pradesh and was confined largely to it. |
| Nagaland, Manipur, Arunachal Pradesh, Assam | The Act of Chapter Five, in force by six-month declaration into 2026. Mon, December 2021: fourteen dead, sanction to prosecute refused. |
| Mizoram, Tripura, Meghalaya | The Act withdrawn — 1980s, 2015 and 2018. Chapter Eight. |
| Gujarat and the Narmada valley | The most documented single displacement in India, and the one that produced a national movement and a Supreme Court case. |
The overlap in this chapter is unusually easy to verify, and that is worth saying because the claim sounds like the kind of thing that would be contested.
Map one is a list. Scheduled Areas are notified by presidential order, district by district, and published.
Map two is a list. Mineral reserves and production are published annually by state and district in official yearbooks.
Map three is a list. The districts classified as affected by left-wing extremism are named in Indian government documents and revised periodically.
Anybody can lay the three lists side by side in an afternoon. The overlap is not an interpretation and it is not disputed.
What none of it can show: causation. Three lists that coincide establish coincidence. The sequence in section 6.2 is what carries the argument, and the sequence rests on dates — 1950 for the designations, the 1950s for the projects, the late 1970s for the insurgency — each of which is independently documented.
What this chapter deliberately does not claim: that extraction produced the insurgency. Section 6.3 establishes an outcome and the hidden assumption box states the limit explicitly.
Which is where the standard account of these districts goes wrong, and it goes wrong in the same way from both directions.
Everybody who discusses the security presence in these districts assumes that the insurgency explains it.
The government’s version: an armed movement required a response. The sympathetic version: a brutal response radicalised a population. Both take the insurgency as the starting point and argue about what followed it.
The insurgency is the last of the three maps to be drawn.
The minerals were there before anybody. The areas were scheduled in 1950 precisely because the people living there were identified as needing protection. The mines, the dams and the steel plants arrived through the 1950s and 1960s. The insurgency became significant in this belt from the late 1970s.
Which reverses the causal arrow that both sides use. The heavy state presence in these districts is not a response to an armed movement — it preceded the armed movement by two to three decades, and what it consisted of was extraction.
This does not excuse the insurgency, and this part is not offering it as one. An armed movement that kills teachers, blows up buses and executes suspected informers is doing those things and Chapter Nine says so. Nor does it establish that grievance caused it; plenty of equally extracted regions produced no insurgency at all, and Part Ten made exactly this point about Punjab.
What it establishes is narrower and harder to dismiss. The question “why is there so much armed force in these districts” has an answer that begins in 1951 and not in 1980, and any account that starts with the insurgency has removed thirty years of the explanation before beginning.
And it explains why the standard prescription fails. Development is offered as the answer to the insurgency. But these are the districts that received development first, in its heaviest form, before almost anywhere else in India — and section 6.3 shows what they have to show for it.
Three lists, one sequence, and one outcome after seventy years.
Three maps — Scheduled Areas, the mineral belt, and the insurgency — overlap almost exactly, and the Indian government’s own documents say so.
The order matters. Minerals always; scheduled in 1950; extraction from the 1950s; insurgency from the late 1970s. The heavy state presence preceded the armed movement by two to three decades.
Seventy years of supplying India’s coal, iron and electricity has left these among the poorest districts in India. That is not a cost and a benefit. It is one end of a pipe.
7The Counting
Part Two’s rule for the fifth time. This case is different from the other four, and the difference produces a sharper version of the rule.
7.1 — The rule, and the four before this
Part Two established a test this series has used ever since. A state that counts everything — its cattle, its rainfall, its literates, its telephone lines — does not fail to count something by accident. Where a count is missing, somebody decided it would be.
Four applications so far, and each had a different mechanism.
The count never made. Part Two: no official enumeration of the dead of Partition was ever attempted, and the range historians use spans a factor of five.
The count made and withdrawn. Part Nine: the Shah Commission counted the detentions, the sterilisations and the deaths, and its report was recalled after the party it examined returned to office in 1980.
The count made by somebody else. Part Ten: a private citizen assembled evidence of secret cremations from firewood receipts because no official record existed. He was abducted and killed. A state agency later confirmed his method in one district.
The count made and not acted on. Part Eleven: 2,733 deaths established by an official committee, 587 police reports, and twenty-five convictions.
7.2 — What is not counted here
Now the list for this part.
How many people the Indian state has displaced. No official national figure. The number everybody uses is a researcher’s.
How much land protected by the Fifth Schedule has been alienated. No national figure.
How much has been restored. No national figure.
How many people have been killed in operations in areas under the Act of Chapter Five, in sixty-seven years. No consolidated figure.
How many claims under the Forest Rights Act have been filed, granted and rejected, and on what grounds. This one is partially available and is the exception that shows the rest is a choice.
| Part | What was not counted | Mechanism |
|---|---|---|
| Two | The dead of Partition | Never made. No enumeration was attempted; the range historians use spans a factor of five. |
| Nine | Detentions, sterilisations and deaths in the Emergency | Made and withdrawn. The report was recalled after the party it examined returned to office. |
| Ten | Secret cremations in Punjab | Made by somebody else. A private citizen assembled it from firewood receipts and was killed for it. |
| Eleven | The dead of November 1984 | Made and not acted on. 2,733 established, 587 reports, 25 convictions. |
| Thirteen | The displaced, the land, the operational deaths | Made everywhere and never added up. Every component sits in a district file. |
7.3 — Why this case is different
In the four earlier instances there was a moment at which a count could have been made and was not, or was made and suppressed. Somebody decided.
Here, the counts exist.
Every acquisition has a file. Every project has a figure for the number of families to be displaced, because the figure is needed to calculate compensation. Every state revenue department has records of restoration proceedings. Every declaration under the Act of Chapter Five is a gazette notification with a date.
The material is not missing. It has never been added up.
Which is a different thing from a missing count, and it is worth naming precisely, because it is the form the problem takes in a modern administrative state that no longer suppresses documents.
Aggregation: combining figures that already exist into a total.
It is worth separating from the words it gets confused with. A census counts things nobody has counted. A survey samples and estimates. An audit checks whether recorded figures are true.
Aggregation does none of those. It asks holders of existing records to report a number they already have, and adds the column.
Why it matters here: the distinction decides whether the absence of a figure is a resource problem or a choice. A census is expensive. A survey needs design and fieldwork. An aggregation needs a circular and a spreadsheet — which is why Chapter Seven’s finding is not that India lacks capacity.
All of which changes what kind of absence this is.
Everybody who writes about this — including the researchers who produced the estimate in Chapter Four — assumes that nobody knows how many were displaced.
It is stated as an information problem. The data is poor, the records are scattered, the definitions are inconsistent, capacity is limited. The implied remedy is a better survey.
Somebody knows every single one of these numbers.
A collector knows how many families were displaced by the project in his district, because he signed the awards. A revenue officer knows how many restoration applications were filed and how many succeeded. A home department knows how many declarations are in force and where.
What has never been done is the addition.
And addition is the cheapest operation in government. It requires no survey, no fieldwork, no new capacity — only an instruction from one office to every district to send a figure it already holds, and a clerk to total the column. India does this routinely for crop yields, school enrolment, electricity connections and vaccination coverage, at far greater cost and difficulty than this would take.
So the honest description is not that the information is unavailable. It is that the aggregation has not been commissioned, for seventy-five years, by any government of any party, while equivalent aggregations were commissioned for hundreds of less consequential things.
Which sharpens Part Two’s rule rather than merely repeating it. The original form was about a count somebody decided not to make. This is the modern form, and it is harder to see and harder to answer: an unmade addition, where every component exists, nothing has been concealed, and no decision was ever taken that anybody has to defend.
Nobody suppressed anything. There is no order to point at. There is simply a column that has never been totalled, and a figure that consequently does not exist — and a figure that does not exist cannot be the subject of a policy, a debate, a demand or an apology.
Which completes the taxonomy, and gives this series its fifth and least answerable form of the rule.
Part Two’s rule, five times now, with five mechanisms: never made; made and withdrawn; made by a private citizen instead; made and not acted on; and now made everywhere and never added up.
No national figure exists for people displaced, protected land alienated, protected land restored, or deaths in operations in sixty-seven years.
Every component exists in a district file. What has never been done is the addition — the cheapest operation in government, commissioned routinely for crop yields and school enrolment.
8What Was Built
There are real successes here, one of them unique in Indian history. They have a feature in common, and the feature is the argument of this whole part.
8.1 — The Sixth Schedule councils
The Autonomous District Councils of Chapter Two are the most substantial devolution in the Indian constitution and they have largely worked as instruments of control over land.
In the Sixth Schedule areas, land alienation on the scale described in Chapter Three did not occur. The reason is structural: a council with legislative power over land, elected by the population the power protects, is a body inside the enforcement chain whose interest runs the right way.
They have their own failures — capture, patronage, disputes with state governments over funds — and they have not delivered development at any impressive rate. What they delivered was the thing the Fifth Schedule promised and did not: the land is still there.
8.2 — Mizoram, and the three withdrawals
Part Twelve’s Chapter Seven recorded the Mizoram Accord of June 1986 as the single most successful thing that government did. It belongs here too, from the other side.
A twenty-year insurgency ended by agreement. The insurgent leadership entered electoral politics. Statehood was granted. Forty years later Mizoram is among India’s most peaceful states and has among its highest literacy rates.
And the Act of Chapter Five was withdrawn — from Mizoram in the 1980s, from Tripura in 2015, from Meghalaya in 2018.
Those withdrawals matter more than they are usually given credit for, because they establish that the declaration can be revoked. It is not a ratchet. It has been reversed three times, which removes the defence that reversal is impossible and leaves only the question of why it has not happened elsewhere.
8.3 — The Forest Rights Act
The 2006 Act is the most serious attempt any Indian government has made at the underlying problem, which Chapter Three identified as the absence of recorded title.
It does three things that nothing before it did. It recognises rights that were never entered on any record. It vests the determination in the gram sabha rather than the forest department. And it states in its own preamble that an injustice was done.
Implementation has been uneven and contested — claims rejected in bulk, community rights recognised far less often than individual ones, and continuing pressure on the consent provisions. But titles have been issued at a scale that is not trivial, and each one converts a person from an encroacher into an owner.
8.4 — Niyamgiri
In 2013 the Supreme Court, in the case concerning bauxite mining on the Niyamgiri hills in Odisha, did something no Indian court had done before.
It did not decide whether the mine should proceed. It directed that the gram sabhas of the affected villages should decide whether the project affected their religious and cultural rights, and that their decision would govern.
Twelve gram sabhas met. All twelve voted against.
The project did not proceed.
That is, as far as this part can establish, the only occasion in Indian history on which the people who would be displaced by a major extractive project decided whether it would happen, and decided no.
Twelve village assemblies met on hillsides in Odisha over several weeks in 2013, under the direction of the Supreme Court, and voted on whether a bauxite mine would be dug above them.
Each one voted no. The mine was not dug.
Set that against everything in Chapters Three to Six — the mechanisms, the sixty million, the ratio, the unmade addition — and it is a single project in a single district.
It is also the only time it has ever happened, which is why it is in this part twice and why Chapter Nine’s rules are built around it rather than around anything larger.
8.5 — What they have in common
Look at the four.
An elected council whose electorate is the protected population. An accord negotiated with the community’s own leadership. A statute that vests determination in the assembly of the affected village. And a court order that handed a decision to twelve such assemblies.
Every one of them works by putting the decision inside the protected population, or by putting the enforcement outside the state government that wants the land.
Which is Chapter Two’s hidden assumption stated as a positive. The protections that worked are the ones that were not handed to the party with the contrary interest — and the record shows this is not a theory, because when the arrangement was built the right way round, it held.
The successes in this chapter suggest an obvious remedy, and the objection to it is serious.
Niyamgiri stopped one project. Extend that rule to every mine, dam, road and transmission line in India and nothing gets built, because there is always somebody who would rather it were built elsewhere. Infrastructure by definition imposes local costs for dispersed benefits, and a system in which each affected village decides will produce a country with no reservoirs and no grid. A national interest that any village can veto is not a national interest. And the beneficiaries of the veto are not only villagers — competitors, incumbents and rent-seekers learn very fast to fund a local objection.
The consent requirement already exists in Indian law for Scheduled Areas specifically — in PESA, in the Forest Rights Act, in the 2013 acquisition statute — because Parliament decided these areas warrant it, having watched what happened for sixty years without it. The question is not whether to create the rule but whether to enforce the one that exists. And the veto objection is answered by the record: consent provisions have been in force for decades and the projects have overwhelmingly proceeded. What consent changes is not whether but on what terms, which is exactly the thing that has never been negotiated.
What would settle it: comparing project completion rates in Scheduled Areas before and after the consent provisions took effect. The data exists in project files and has not been assembled, which is Chapter Seven again.
Why people care so much: because this is the live argument in Indian development policy right now, and both sides believe the other is arguing in bad faith about who benefits.
The credit column for this part, and the single feature that runs through all of it.
Four real successes: the Sixth Schedule councils, where the land is still there; Mizoram; the Forest Rights Act; and Niyamgiri, where twelve gram sabhas voted and the project stopped.
The Act of Chapter Five has been withdrawn three times — Mizoram, Tripura, Meghalaya. Which establishes that the declaration is not a ratchet and can be reversed.
What they have in common: each puts the decision inside the protected population or the enforcement outside the state that wants the land. When the arrangement was built the right way round, it held.
9What Would Have Prevented It
The defence includes something this part must concede outright: the insurgency’s own record. Then five rules, all of them procedural.
9.1 — Four arguments for the defence
These are the strongest available and the third is conceded without qualification.
One. The projects were necessary and there was no capital. Part Four described the model and its costs, and its premise stands: a country with no power, no steel and no irrigation had to build them, and could only build them where the coal and the river valleys were. The alternative to Hirakud was not a better Hirakud.
Two. The formal protections are among the strongest anywhere. India wrote two Schedules into its constitution in 1950 to protect indigenous land, at a time when comparable countries were doing nothing of the kind and several were still actively dispossessing. The Sixth Schedule is more generous than almost any equivalent arrangement in the world. That is a real achievement and Chapter Eight shows it working where it was structured correctly.
Three. The insurgency has its own record and this part concedes it. The Maoist movement in these districts has killed civilians in large numbers, executed people as suspected informers after proceedings that were not trials, destroyed schools, blown up buses and killed elected representatives. Its presence has also directly obstructed the arrival of the delivering half of the state — the schools and clinics whose absence Chapter One describes. Nothing in this part is an argument that the armed movement is a legitimate remedy, and a reader who takes it that way has read it wrong.
Four. The successes are real. Chapter Eight, in full.
It is worth putting a price on the rules that follow, because the usual objection to reform in this area is cost.
Rule three — the national return — is a circular to every district collector asking for figures already in their files, and a section in an existing annual report to publish the total. One officer, part-time.
Rule four — parliamentary renewal of declarations — is a debate every six months on a motion that already has to be signed by a minister.
Rule five — sanction decided by a court — adds a hearing to a decision currently taken in a file.
Against which set what the absence has cost: seventy-five years, sixty million people, and a figure that still does not exist.
Nothing in Chapter Nine failed on expense. The rules are cheap, and that is the most damaging thing this part can say about their absence.
9.2 — Five rules
Each of these was available, none requires a constitutional amendment, and none is expensive.
One. Standing for the protected party. The gram sabha, or a community body, should be able to enforce Fifth Schedule land regulations directly in court, without depending on the revenue administration of section 2.4. The single change that would have done most.
Two. Consent recorded by somebody independent of the acquirer. The consent provisions exist. What does not exist is verification by anybody other than the district administration that wants the project.
Three. A statutory duty to compile and publish. An annual national return of displacement, land alienated and land restored, aggregated from figures the districts already hold. Chapter Seven’s whole argument reduces to this one line.
Four. A real sunset on disturbed-area declarations. Not a six-month renewal by notification, but expiry unless affirmatively renewed by Parliament, with reasons stated and laid before it. Part Nine recorded exactly this reform being made for national emergencies in 1978 — the mechanism exists and was simply never applied here.
Five. Sanction decided by a court, within a fixed period. Section 6's protection can be preserved while removing the feature that makes it absolute: let a court decide whether the acts were done in the course of duty, on a timetable, with reasons. Refusal without reasons and without limit is what converts immunity from a safeguard into a bar.
Section 9.2's rules have an unusual property for this series: four of the five already exist somewhere in Indian law, which is how we know they are workable.
Standing for an affected party to enforce a protection directly is ordinary Indian public interest practice and is how Samatha and Niyamgiri reached court at all.
Independent verification of consent is what the Supreme Court ordered in Niyamgiri and it was carried out in weeks.
Statutory duties to compile and publish annual returns exist for dozens of subjects.
Parliamentary sunset with affirmative renewal was built into the emergency provisions in 1978, as Part Nine recorded.
Only the fifth — judicial determination of sanction — would be new, and it is a narrower change than the 2016 judgment already required.
What this cannot show: that any of them would have worked at scale. Each has been demonstrated in a single case or a neighbouring field, and a rule that works once is not a rule that works everywhere.
9.3 — What I would have to retract
Three things would break this part.
If the aggregation in rule three were done and showed displacement far below sixty million, with the Scheduled Tribe share near the population share, then Chapter Four and much of Chapter Six collapse.
If the declarations under the Act of Chapter Five were shown to rest on assessments that are genuinely prepared, reviewed and tested every six months, then Chapter Five’s central claim — that renewal is a formality — fails.
If forest rights titles turn out to have been issued at scale and to have held against acquisition, then Chapter Two’s finding about the enforcement architecture is weaker than stated, because the architecture will have corrected itself.
The first depends on an addition nobody has done. The second on files the government holds. The third on data being generated now.
Four defences: the projects were necessary; the formal protections are among the strongest anywhere; the insurgency’s own record is conceded outright; and Chapter Eight’s successes are real.
Five rules: standing for the protected party; consent verified by somebody independent of the acquirer; a statutory duty to compile and publish; a real parliamentary sunset on declarations; and sanction decided by a court on a timetable.
None requires a constitutional amendment. The sunset mechanism already exists — Part Nine recorded it being built for national emergencies in 1978 and it was never applied here.
10An Honest List of What We Do Not Know
This is the shortest list of solid facts in the series and the longest list of missing ones, and the imbalance is the finding rather than a defect of the research.
10.1 — Genuinely unknown
Six things, and every one of them is a number the Indian state could produce.
How many people have been displaced. The working figure is more than sixty million and it is a researcher’s estimate.
Why it is unknown: the addition has never been commissioned. Chapter Seven.
How much Fifth Schedule land has been alienated, and how much restored.
Why it is unknown: state revenue records hold both figures and no national compilation exists.
How many people have been killed in operations under the Act of Chapter Five since 1958.
Why it is unknown: no consolidated return is published. Individual incidents are documented and the total is not.
What the criteria for a disturbed-area declaration actually are.
Why it is unknown: none are stated in the Act and none have been published. The declarations give no reasons.
How many Forest Rights Act claims have been rejected and on what grounds. Partially available, unevenly by state.
Why it is unknown: reporting is not standardised and rejection grounds are frequently not recorded.
Where the output went. The front matter’s argument box named the study that would settle this part’s central question — a project-by-project comparison of where the electricity, steel and irrigation ended up against where the displaced people ended up.
Why it is unknown: nobody has done it. The material is in project files.
10.2 — Solid
The law. The Fifth and Sixth Schedules, PESA, the Forest Rights Act, the Act of Chapter Five with its 1972 amendment and section 6, and the notifications currently in force. All published.
The judgments. Naga People’s Movement (1997), Samatha (1997), the Niyamgiri decision (2013) and the 2016 holding that there is no blanket immunity. All reported.
The inquiries. The Jeevan Reddy Committee’s recommendation of repeal in 2005 and that it was not acted on. The Santosh Hegde Committee’s finding in 2013 about the sample it examined.
Mon, December 2021. The deaths, the state cabinet’s recommendation, the state’s attempt to prosecute, and the refusal of sanction.
The three withdrawals — Mizoram, Tripura, Meghalaya.
Niyamgiri. Twelve gram sabhas, all against, project stopped.
The census. Scheduled Tribe population, share and distribution.
The overlap of the three maps. Stated in Indian government documents.
10.3 — The addition nobody has done
One observation to close on, because it is the thing I did not expect when I started this part.
In every earlier part of this series, the missing evidence was behind something. A closed cabinet record. A recalled report. A file the government holds and will not open. In each case there is a door and somebody is standing at it, and the sentence “this depends on files the Indian government holds” has now ended seven consecutive parts.
Here there is no door.
Nothing in Chapter Seven’s list is classified. No official has refused to release it. There is no case pending for its disclosure and no minister has declined to answer. Every component sits in a district office in an open file, and the figures are collected, filed and superseded year after year without anybody ever being asked to total them.
Which is why this is the part I would most like to be wrong about, and the one where being wrong is easiest to establish. Any government that wanted to could produce every figure in section 10.1 within a year, at negligible cost, using powers it uses for far smaller purposes every month.
None has. And unlike every other gap in this series, no one has ever had to decide not to.
Genuinely unknown: how many were displaced, how much protected land moved and how much came back, how many have been killed under the Act, what the declaration criteria are, how many forest claims were rejected, and where the output went.
Solid: the law, the judgments, the inquiries, Mon, the three withdrawals, Niyamgiri, the census, and the overlap of the three maps.
In every earlier part the missing evidence was behind a closed door. Here there is no door — and no one has ever had to decide not to open it.
Sources & further reading — Part 13
Timeline
Dates only. The arguments are in the chapters.
| Date | What happened |
|---|---|
| 15 Aug 1942 | The colonial Armed Forces Special Powers Ordinance, to suppress the Quit India movement. The ancestor of the Act of Chapter Five. |
| 1894 | The Land Acquisition Act, requiring no consent from anybody. It governs acquisition until 2013. |
| 1950 | The constitution comes into force with the Fifth and Sixth Schedules. Scheduled Areas are designated in what become ten states. |
| 1950s | The first great projects — Damodar Valley, Hirakud, Rourkela, Bhilai — are built in these districts. Part Four’s economic model. |
| 1955 | Assam enacts a Disturbed Areas Act. |
| 11 Sep 1958 | The Armed Forces (Special Powers) Act replaces an ordinance of May that year. |
| 1967 | The movement that becomes the Maoist insurgency begins in Bengal. It becomes significant in the mineral belt from the late 1970s. |
| 1972 | The Act is amended so the centre may declare any area disturbed without the state’s consent. |
| 1980s | The Act is withdrawn from Mizoram. |
| 30 Jun 1986 | The Mizoram Accord ends a twenty-year insurgency. Part Twelve’s Chapter Seven. |
| 1997 | Naga People’s Movement of Human Rights: the Supreme Court upholds the Act’s validity with safeguards. Nagaland’s principal groups enter ceasefire in the same year. |
| 1997 | Samatha: government land in Scheduled Areas cannot be leased to non-tribals for mining. Application confined largely to one state. |
| 24 Dec 1996 | PESA extends panchayats to Scheduled Areas and gives gram sabhas power to prevent alienation and restore land. Enacted under a United Front government. |
| 2000–2016 | Irom Sharmila’s hunger strike in Manipur demanding repeal. Sixteen years. |
| 2005 | The Jeevan Reddy Committee, appointed by the government, recommends repeal. The report is not acted on. |
| 2006 | The Forest Rights Act. Its preamble admits rights were never recorded and that a historical injustice resulted. |
| 2013 | The Santosh Hegde Committee finds that none of the Manipur encounters in its sample was genuine. |
| 2013 | Niyamgiri. The Supreme Court directs the gram sabhas to decide. All twelve vote against. The project stops. |
| 2013 | The 1894 Act is replaced by a statute with consent and impact-assessment requirements. |
| 2015, 2018 | The Act is withdrawn from Tripura, then from Meghalaya. |
| Dec 2021 | Mon, Nagaland. Fourteen civilians dead. The state cabinet recommends repeal; the state seeks to prosecute; sanction is refused under section 6. |
| 2026 | Declarations remain in force in parts of Nagaland, Manipur, Arunachal Pradesh and Assam. Sixty-seven years. |
Part Fourteen starts here. It returns to the chronology at the point Part Twelve left it — a party out of office for the first time since 1977, and about to come back under a Prime Minister nobody expected.
Glossary
Every hard word used in this part, in plain English.
| Word | What it means |
|---|---|
| Adivasi | “Original inhabitant.” A political term preferred by many of the communities the constitution calls Scheduled Tribes, and contested precisely because of the claim it makes. |
| Autonomous District Council | An elected body under the Sixth Schedule with actual legislative power over land, forests, inheritance and village administration. The most substantial devolution in the constitution. |
| Disturbed area | A declaration that brings the Act of Chapter Five into force. Made by six-month notification, on no published criteria, and since 1972 without needing the state’s consent. |
| Eminent domain | The power of a state to take private property for public purposes on payment of compensation. Under the 1894 Act it required no consent from the owner. |
| Gram sabha | The assembly of every adult in a village, as opposed to the elected panchayat, which is a committee. The hardest body in Indian public life to buy, which is why the consent laws vest their powers in it. |
| Minor forest produce | Everything a forest yields other than timber — leaves, resin, honey, fruit, bamboo. A principal source of income, owned by nobody on the record, and therefore valued at nothing when land is acquired. |
| PESA | The Panchayats (Extension to Scheduled Areas) Act, 1996. Gives gram sabhas power to prevent alienation of tribal land and to restore land unlawfully alienated. |
| Sanction to prosecute | Permission that must be obtained before a public servant is tried for an official act. Under section 6 it is given by the central government, with no time limit, no duty to give reasons and no effective appeal. |
| Scheduled Area | Territory designated under the Fifth Schedule, in ten states, where the Governor may prohibit transfer of tribal land. |
| Scheduled Tribe | A legal category: communities listed by presidential order under Article 342, entitled to reservation and to specific land protections. Around 8.5 per cent of India’s population. |
What Comes Next
Part Fourteen — 1991
Part Fourteen returns to the chronology. A party out of office, a Prime Minister nobody expected, and the year the economic model of Part Four was dismantled. It asks:
- Part Four established what the licensing system did and who it served. What exactly was dismantled in 1991, by whom, and how much of it survived?
- The reforms were introduced by a minority government with 244 seats. Part Twelve found that 414 bought very little. What does that comparison establish?
- Was 1991 a decision or a surrender? The balance of payments position, the gold, and what the conditions actually were.
- Who won and who lost, on Part Five’s method — measured by region, sector and class rather than by aggregate growth?
- December 1992 happened under this government. What did it do, what did it know, and what did the commission find?
- Part Twelve found the failures structural rather than personal. Does a Prime Minister with no majority and no family claim change the finding?
Questions, not topics. If any of them is answered badly, you will be able to tell.