Where We Left Off
This part covers mass killing in which every community was both victim and perpetrator. Where a figure is disputed you get the range and the method behind each end of it. Where no count was ever taken, that silence is treated as a finding.
Part One built the tools. This part is the first time they get used on anything.
Part One was about method, and it made four commitments that this part now has to keep.
The first was that this series selects like a prosecutor and checks like an auditor. The choosing is one-sided and says so. The evidence is not. That distinction is the whole spine of the project, and the reason for it was strategic rather than polite: the true record of Congress in power is damaging enough on its own, and every invented story added to it hands the party an easy escape from all the real ones.
The second was the grading scale. Every serious claim carries a label — convicted, court-found, commission-found, audited, alleged, or fabricated — printed where you can see it. You will meet all six in this part, including a case where a government’s own commission found something and the government then made the finding disappear.
The third was that a charge needs an alternative. Saying a government blundered means claiming there was another road, that it could be seen from where they stood, and that somebody could have taken it. Where I have all three, the charge is heavy. Where I have fewer, I use a smaller word.
The fourth was the rule that will do more work in this part than anywhere else in the series: when a state that counts everything has no count of something, somebody decided that. The missing number is a finding, not a gap. This part contains one of the largest missing numbers in the history of the Indian Union, and a commissioned report that went missing along with it.
Part One also declared where I stand and I will do it again here, more narrowly, because this part is where it bites. My previous work was a fourteen-part history of Punjab. That means I arrive at the chapters on 1947 with attachments, and Punjab in 1947 is a place where every community on the ground was both victim and killer. If I find myself writing one of them as only the first thing, I have stopped doing history. Watch for it. I will flag it where I catch it myself.
How to read this book
The Six Boxes
Six coloured boxes run through every part of this series. Each does one job. Here they are with live examples from the material ahead, so you learn the notation before you need it.
A Word Box explains a hard word the moment it first appears.
Paramountcy: the arrangement under which the British Crown controlled the foreign affairs and defence of India’s princely states while leaving their rulers in charge internally. The states were not British territory. They were tied to Britain by treaty.
When the British left, paramountcy lapsed — it simply ended, and it was not automatically transferred to India or Pakistan. In law, every one of those states became free-standing on 15 August 1947.
Why it matters here: roughly two-fifths of the land area of British India was not British India. Chapters Five to Eight are about what happened to it.
An In Real Terms box turns a number too big to picture into something with a body.
Estimates of how many people were displaced by Partition run from about 12 million to about 20 million.
Take the lower figure. That is roughly the entire population of a large modern country picking up what it can carry and walking, in about eighteen months, in two directions at once, along roads that were also being used by the people going the other way.
If you lined up 14 million people in single file, a metre apart, the line would go round the Earth’s equator about a third of the way. They were not in a line. They were in columns of tens of thousands, on foot, in monsoon heat, being attacked.
A How We Actually Know This box shows the physical evidence and then says what it cannot prove.
Much of what is known about refugee movement in 1947 and 1948 survives because refugees had to be fed and housed. Camp registers, ration issues, railway movement orders and resettlement claims all exist as administrative by-products.
This means the arrivals are far better recorded than the departures, and the survivors far better recorded than the dead. A person who reached a camp entered the record by the act of surviving. A person killed on the road entered nothing.
What it cannot show: the size of the gap between those two groups. That is the central measurement problem of this entire part, and Chapter Three is about it.
An Argument box appears where serious people disagree. Each side gets its strongest case, not a weak version built to be knocked over.
Everything in this part follows from the division of the country. So before anything else: was accepting it a blunder?
Congress accepted a plan in June 1947 that it had rejected in principle for years, under a deadline set by a departing power that had no authority to impose one. It accepted a compressed timetable — independence brought forward by nearly a year — with no boundary agreed, no population transfer planned and no force in place to protect anybody. The leadership was exhausted and wanted power. That is a reason, not a justification.
By 1946 the country was already killing itself. Calcutta in August 1946 left about four thousand dead. The alternative on the table was a weak centre with permanent communal vetoes, which would have produced either paralysis or the same violence spread over decades. And Congress did not control the timetable; the British did, and the British were leaving whether or not anybody was ready.
What would settle it: nothing. It is the largest counterfactual in modern Indian history and Part One explained why those cannot be measured.
Why people care so much: because whoever is blamed for Partition is blamed for everything downstream of it, which is most of the last seventy-nine years, including two nuclear weapons programmes.
A Hidden Assumption box digs out a belief sitting underneath an argument that both sides accept without discussing.
Everybody arguing about 1947 — every side, every country — treats Partition as one event.
It was at least three, with different agents and different timing. There was a constitutional division, agreed in June and enacted in July, which created two dominions. There was a boundary award, drawn by one exhausted Englishman in about five weeks and published two days after independence. And there was a wave of killing and expulsion, which began well before either, in Calcutta in 1946, and continued into 1948.
Running the three together is enormously convenient, and it is convenient in every direction. It lets a British account say the violence was inherent to the place. It lets an Indian account say the violence followed from the division. It lets a Pakistani account say the division followed from the violence. Each is choosing a causal order, and the merging of the three events is what hides the choosing.
Keep them apart while you read this part. When you see a claim about “Partition”, ask which of the three it is about. Most of the arguing dissolves.
And every chapter closes with a Remember This box: the chapter in the plainest words available. A reader who read only those boxes would still have the whole argument.
Part One built the method: one-sided selection, honest evidence, a grade on every claim, and an alternative named before anything is called a blunder.
The rule that matters most here: when a state that counts everything has no count of something, somebody decided that.
“Partition” is three separate things wearing one name — a constitutional division, a boundary award, and a wave of killing. Keep them apart and most of the shouting stops.
1What Was Actually Handed Over
On 15 August 1947 something changed hands. Almost every argument about what followed depends on being precise about what that something was.
1.1 — A machine, not a country
Picture the handover as most people picture it. A flag comes down, another goes up, and a country that was being held by somebody else is returned to the people who live in it.
That picture is wrong in a way that matters for everything in this series.
What changed hands was not a country. It was an administrative machine that had been built over about ninety years to govern a conquered population, and a map that machine had drawn. On the morning of 16 August the machine was still there. The same offices, the same files, the same ranks, the same laws, largely the same men. What had changed was who gave it instructions.
And it was less than the map suggested. British India — the territory the British actually ruled — was not the whole of the subcontinent. Roughly two-fifths of the land area was princely states, tied to the Crown by treaty but not governed by it. Chapters Five to Eight are about them. Set them aside for now and look at what Congress inherited directly.
Dominion: a self-governing country inside the British Commonwealth, with the British monarch still as formal head of state.
India and Pakistan both became dominions in August 1947, not republics. India remained one until 26 January 1950, when the constitution came into force. In the meantime the head of state was still the King, represented by a Governor-General, and the working constitution was still a British statute.
Why it matters here: for the first two and a half years of independence, India was governed under law written in London for the purpose of managing India.
1.2 — What came with the building
The inheritance had five main pieces, and all five turn up later in this series.
A civil service. A small, elite, extremely capable administrative corps that ran districts, collected revenue and reported upwards. By 1947 about half of its officers were Indian. Sardar Patel fought to keep it, against people in his own party who wanted it dissolved as a colonial instrument, and he won. That decision is one of the strongest defences available to the Congress record and it belongs on the page.
An army. Professional, disciplined, and — this is the part that mattered in 1947 — accustomed to staying out of politics. It was split between the two dominions in a matter of weeks, which is itself an administrative feat. Unlike almost every comparable country, India’s army never took power. That is not an accident and it is not luck.
A revenue and police system. Designed to extract money and maintain order in a population assumed to be hostile. It worked well at both.
A legal code. Criminal law, civil procedure, and a body of emergency and preventive powers written for a government that expected to be resisted.
A working constitution. The Government of India Act of 1935, kept on as the interim constitution until January 1950. Chapter Nine is about how much of it survived into the permanent one.
1.3 — What did not come with the building
Now the other column, because a fair account needs both.
There was no experience of running a foreign policy, because there had not been one. No experience of running a central bank in a sovereign state. No industrial base to speak of relative to the size of the population. A literacy rate somewhere around one adult in six. And a set of famine and epidemic vulnerabilities that had killed millions within living memory.
Against that, the new government had one asset that is easy to undervalue because it cannot be counted: legitimacy. The people taking charge had spent decades in a mass movement and many of them had spent years in prison for it. When they told the country to do something, an unusual number of people did it.
Picture inheriting a large, well-built factory. The machines work. The foremen know their jobs. The accounts are in order.
Now notice three things. The factory was built to make one product, and you want it to make a different one. Most of the people who designed it have gone home, and the ones who stayed learned their trade making the old product. And there is no manual for conversion, because nobody has ever converted a factory this size before.
That is the position in August 1947, and it explains both the achievements and the failures in this part. The machine did what machines do. It kept running in the direction it was pointed.
1.4 — The men who stayed
One consequence of keeping the machine deserves stating plainly, because it runs through the next eighteen parts.
An officer who had spent 1942 suppressing the Quit India movement was, in 1948, a District Magistrate in independent India. He had not changed. His powers had not changed. His training had not changed. What had changed was that the instructions now came from the people he had previously been arresting.
This is not a criticism of those officers, most of whom served the new state loyally and some of whom served it heroically. It is a statement about capability. A government that inherits an apparatus built for control will find that the apparatus is very good at control, and that reaching for it is the path of least resistance. When you read, in Chapter Nine, that independent India passed a law allowing detention without trial in its first year, remember that it did not have to invent the idea. It had the drafting on a shelf.
Everyone arguing about 1947 — admirers and critics alike — assumes that what changed hands was a country.
The admirer’s version: a great nation recovered its freedom, and the difficulties afterwards were the difficulties of a young country finding its feet. The critic’s version: a country was handed to people who mismanaged it. Both are arguing about the quality of the management. Both have quietly agreed on what was being managed.
But “India” as a single political unit governed from one capital was itself a British product, assembled by conquest and treaty over about a century, and it did not exist in that form before. What changed hands in August 1947 was the apparatus — a service, an army, a revenue system, a body of law — together with a claim over territory that had never once, in its history, been a single self-governing state.
Why this is not a debating point. If you inherit a country, your job is to govern it. If you inherit an apparatus plus a claim, your job is to make the country, and that is a different job with different methods. It involves persuading or compelling several hundred separate polities to join. It involves deciding what to do about the ones that decline. Chapters Five to Eight are about exactly that work, and they are usually written as though it were housekeeping.
The assumption cuts both ways, which is how you can tell it is doing real work. It lets the defender describe coercion as administration. It lets the critic describe an unprecedented state-building operation as ordinary governing that went badly. Neither has to argue for the frame, because neither has noticed choosing it.
Hold that distinction as you read. This part is not the story of a country being governed badly or well. It is the story of a country being assembled, at speed, by an apparatus designed for something else, while several million people were on the roads.
What changed hands in August 1947 was an administrative machine and a claim, not a finished country. The machine was built to govern a population assumed to be hostile, and it was very good at that.
The inheritance was real and mostly kept: a civil service, an apolitical army, a revenue system, a legal code, and a British statute serving as the constitution until 1950. Keeping the first two was a Congress decision and a good one.
A government that inherits an apparatus built for control does not have to invent the tools of control. It finds the drafting on a shelf. Chapter Nine is what happened next.
2The Line Nobody Was Shown
The border was finished on 12 August and published on 17 August. Millions of people spent the days between them guessing which country they were in.
2.1 — Five weeks
Sir Cyril Radcliffe arrived in Delhi on 8 July 1947. He was a British lawyer of great reputation and no relevant experience. He had never been to India. That was considered a qualification, on the reasoning that a man who knew nothing could not be partial.
He was given the job of chairing two boundary commissions — one for Punjab, one for Bengal — and dividing two of the most densely populated provinces on Earth. Each commission had four other members, two nominated by Congress and two by the Muslim League, and each was a judge. In practice those four deadlocked on every question, which left the decisions to the chairman.
He had about five weeks.
Boundary award: the formal published decision of a boundary commission, fixing exactly where a new border runs.
The word “award” is borrowed from arbitration. It means a decision the parties have agreed in advance to accept, whether or not they like it. Congress and the Muslim League both gave that undertaking before the work began, which is why neither could reject the result afterwards, and both complained bitterly about it anyway.
Why it matters here: because everyone had pre-agreed to accept it, the only real lever left was when it was published.
2.2 — Finished on the twelfth
Here is the sequence, and it is not disputed by anybody.
Radcliffe finished the work on 12 August 1947. He submitted the sealed award to the Viceroy on 13 August. Mountbatten did not publish it. He went to Karachi for the Pakistani independence celebrations on the 14th, and spoke to the Indian Constituent Assembly on the 15th as India’s first Governor-General, and said nothing about it in either place. On 16 August he distributed copies to the Indian and Pakistani leaderships in a meeting. Both sides objected furiously to parts of it, and both agreed to let it stand. It was published on 17 August 1947.
So for the five days across which two countries came into existence, nobody living near the border knew which one they were in.
Imagine your town is going to be in one of two new countries and the decision has already been made, written down, and locked in a drawer four hundred miles away. You are told it will be announced in a few days. You are also being told, by people around you, that men from the other community are arming.
You have two choices. Wait for the announcement, which might place you on the wrong side with everything you own. Or act now on your guess.
Several million people made that calculation in the same week. That is what the delay actually was: not an administrative detail, but a period in which the only information available about the most important question in your life was rumour.
2.3 — Who decided the delay, and why it matters
This is where a careless prosecution brief would overreach, so let me be exact.
The decision to withhold the award was Mountbatten’s. He was the Viceroy, it was his to publish, and he held it. The reasons given at the time were that announcing a border before the transfer of power would create chaos on the day, might wreck the independence ceremonies in both capitals, and would leave Britain announcing a contested line while it was still responsible for order.
V.P. Menon, an Indian official and one of the most important figures in this whole part, advised him that releasing it before 15 August without consulting the two leaderships was a bad idea.
And on 16 August, with the award in front of them, the Indian and Pakistani leaders agreed to publish it as it stood rather than fight over it in public.
So the honest verdict is not “Congress hid the border”. It is narrower and it still holds: the timing was a British decision, taken on Indian advice, and accepted by Indian leaders who had the document in their hands the day before it was published and did not demand it go out immediately. Nobody in that room appears to have argued that people on the ground needed to know first.
Almost everything terrible in the next chapter happened in places whose future was unknown to the people living in them. So: was the delay a mistake, or the least bad option?
The killing in Punjab was already under way before 12 August and had been building for months; it did not need a map to start. Publishing a contested line while the British were still legally responsible for order, with the Punjab Boundary Force already visibly failing, would have handed both sides a grievance and a target on the same day. And the four days changed nothing that mattered: the same line, the same people, four days later.
Publication was the one lever left after both parties had pre-agreed to accept the result. Used differently — announced in July, with the award drawn earlier and movement organised around it — it could have converted a panic into a process. Instead the most consequential fact in millions of lives was withheld precisely during the days when knowing it might have let people leave in order rather than in flight. The reasons given are all about ceremonies and appearances.
What would settle it: a comparison with an area where the boundary was known in advance and movement happened anyway. Nobody has assembled one, and the material to do it exists in district records.
Why people care so much: because if the delay mattered, responsibility sits with a departing British Viceroy, which is comfortable in Delhi and Islamabad alike. If it did not, responsibility shifts to whoever organised the killing, which is uncomfortable everywhere.
Before leaving the timing, it is worth pausing on how firmly any of this can be established, because the rest of the part leans on documents of very uneven quality.
The chronology in this chapter is unusually secure, and it is worth knowing why. It rests on dated official papers rather than on recollection: the award itself carries its completion date, the transfer-of-power records show when it went to the Viceroy, and the meeting of 16 August is minuted. British, Indian and Pakistani accounts, which disagree about almost everything else in 1947, do not disagree about these dates.
When sources with opposing interests agree on a fact, the fact is usually safe. That is one of the few reliable tests available to anybody reading history without access to archives.
What it cannot show: what was said in the room. Minutes record decisions, not arguments. Whether anybody present on 16 August urged immediate publication, and was overruled, is exactly the sort of thing that minutes are designed not to preserve.
2.4 — What the line did
One detail from the award tells you what kind of document it was.
The Ferozepur area contained the headworks of canals that watered a large stretch of what became Pakistani Punjab. Radcliffe’s preliminary thinking had placed parts of it on the Pakistani side. The final award placed the headworks in India. The reasoning available for that change is about not cutting an irrigation system in half and about the needs of the princely state of Bikaner, which depended on those flows.
Whatever the reasoning, note what the boundary was doing. It was not only sorting people by religion. It was cutting a canal network, a railway system, a jute economy and a wheat economy, and every cut it made created a bilateral dispute that had to be managed afterwards by two governments that did not trust each other. Water in particular does not accept a line on a map. Part Seven of this series returns to it, and it is still unresolved today.
One British lawyer with no experience of India divided two provinces in about five weeks, with four Indian judges deadlocking beside him so every real decision was his.
The award was finished on 12 August, handed to the leaders on the 16th, published on the 17th. The withholding was Mountbatten’s decision, taken on Indian advice, and Indian leaders held the document a day before publication without demanding it go out.
The delay did not start the killing. It did remove the one thing that might have turned flight into movement. Both sentences are true.
3The Crossing
Somewhere between two hundred thousand and two million people were killed. That range is not a failure of scholarship. It is a finding about what nobody wanted to know.
3.1 — What happened
The pattern was consistent enough across hundreds of miles to be described in one paragraph, which is itself a fact worth noticing.
A village or a district would receive news, usually rumour, of killings elsewhere. Men would arm. Members of the minority community would be given a period to leave, or would not. Columns of refugees formed, sometimes tens of thousands strong, moving on foot. The columns were attacked on the road. Trains carrying refugees were stopped and boarded. Women were killed by their own families to prevent capture, and abducted in very large numbers by attackers. Property changed hands within days and was occupied by the incoming refugees moving the other way.
Every community did this. Muslims did it to Hindus and Sikhs; Hindus and Sikhs did it to Muslims. In Punjab, where the killing was worst, all three were both. This is the point at which I said in the front matter that I would watch myself, so let me be blunt about it: any account of Punjab in 1947 that reads as though one community was principally acting and the others principally suffering is a political document, not a historical one, whichever community it favours.
Communal: in Indian usage, relating to conflict between religious communities. “Communal violence” means violence between Hindus, Muslims, Sikhs or others as groups.
Notice what the word does. It names the participants and says nothing about organisation. A phrase like “communal violence broke out” describes an event with no author — like weather. Part One warned about exactly this kind of word, and this part is where it earns the warning.
Why it matters here: where evidence shows organisation — lists, transport, coordinated timing, participation by men in uniform — this series will say organisation, and where it does not, it will not.
3.2 — The numbers, and the size of the disagreement
Now the arithmetic, which is where most accounts go soft.
Published estimates of the death toll range from about 200,000 to about 2 million. Displacement estimates run from about 12 million to about 20 million. As many as 100,000 women are estimated to have been raped or abducted.
The death range is a factor of ten. Part One said that a spread that wide is almost always a difference of method rather than of data quality, so here are the actual positions, which are more informative than the range.
| Estimate | Source and approach |
|---|---|
| ~180,000 | A contemporary British official estimate, built from administrative reports reaching the government. |
| ~500,000 | A contemporary Indian estimate, split roughly evenly between Muslims and non-Muslims. |
| 200,000–360,000 | The range one scholarly review considers closest to the truth, while noting that a wider “consensus” figure of around 500,000 circulates. |
| 400,000 ±100,000 | A demographic estimate, reasoning from population data rather than reports. |
| ~1,000,000 | A widely cited figure from oral-history-based work, again split roughly evenly. |
| up to 2,000,000 | The upper end of the published range. |
Look at what separates them. The low figures come from reports that reached an administration. The high figures come from demography and testimony. The low figures count deaths somebody wrote down. The high figures try to count deaths that happened.
In a collapse where the administration itself had split in two, where district officers were fleeing alongside everybody else, and where whole villages emptied with no survivor left to report anything, the reporting method has an obvious and predictable bias. It undercounts precisely where the violence was worst, because the worst places produced no reporter.
The demographic approach works differently and is worth understanding, because it is the strongest tool available here. You take the census of 1941, project forward what the population of a district should have been by 1951, count what was actually there in 1951, and subtract. What is left is people who died, were never born, or moved.
Its strength is that it needs no witness. A village that was wiped out completely still leaves a hole in the 1951 count.
What it cannot show: it cannot separate the dead from the migrated. Somebody missing from a district in 1951 might be in a camp four hundred miles away. Untangling those requires assumptions about migration, and those assumptions are where the estimates diverge. It also inherits every flaw of the 1941 census, which was taken in a politically charged atmosphere where community numbers were themselves contested.
3.3 — Delhi, September 1947
One episode deserves separating out, because it happened where the government was sitting.
In September 1947 the violence reached Delhi itself. Muslim neighbourhoods in the capital were attacked, large areas emptied, and a substantial part of the city’s Muslim population either fled to Pakistan or took shelter in camps inside the city, including in the Purana Qila and at Humayun’s Tomb. Property changed hands within weeks, much of it occupied by refugees arriving from the west.
Two things about this are worth holding.
The first is that the government acted, personally and visibly. An emergency committee was formed. Nehru went into the streets himself. Gandhi came to Delhi and fasted, in January 1948, explicitly to stop the killing and to force the payment of money owed to Pakistan, and it worked. Whatever else is in this chapter, the leadership of independent India did not treat the massacre of Muslims in its own capital as somebody else’s problem, and several of them risked their lives over it. Gandhi was assassinated for it within days.
The second is that the police and the administration in the capital did not stop it, and were in places part of it. This is the pattern that recurs at every point in this series where large-scale communal killing happens in an Indian city — 1984, 1992, 2002. The apparatus does not fail randomly. It fails in one direction, and the direction is the same each time.
Delhi in September 1947 is therefore the first entry in a much longer argument, and it is the entry where the political leadership’s own conduct is at its best. That combination — leaders acting, machinery not — is the thing to watch for.
3.4 — The count that was never attempted
Now the finding, and it is the central one of this chapter.
No government on either side ever attempted a count. Not in 1948, when the camps were full of people who could have been asked. Not in 1951, when a census was taken anyway and the questions could have been added. Not since.
Apply the rule from Part One. This was not a state that could not count. It counted crop yields by district. It counted cattle. It counted refugees precisely enough to feed them, because feeding required a number. Within four years it conducted a full national census. The capacity was there and it was used constantly.
So the absence is a decision, and the interesting question is who benefited from it.
The answer is uncomfortable, because it is everybody. A firm number would have had a shape. It would have shown which districts, which weeks, and which direction the killing ran hardest. It would have converted “communal violence” into a map, and a map has authors on it. Two new governments, each needing a founding story and neither able to survive a founding story in which its own majority community was a principal killer, had a shared interest in the number not existing.
Every account of 1947, in India, Pakistan and Britain, shares a premise that none of them argues for: that the violence was something that happened rather than something that was done.
Look at the standard vocabulary. Riots. Communal frenzy. Bloodshed. Carnage. Madness. The subcontinent “descended into” violence. Passions “were inflamed”. Every one of these is a weather word. Weather has no author, no planning meeting and no chain of command.
But the events described were not weather. Columns were ambushed at chosen points. Trains were stopped where stopping them worked. Attacks in different places began within days of each other. Armed bands moved between districts. In some areas men in uniform were among the attackers, and in others local officials were part of the arrangement. Organisation of that kind leaves evidence and some of it survives.
Here is why both sides accept the weather frame. If it was weather, no Indian government has to explain why its own police stood by, and no Pakistani government has to either. The British get to describe an ungovernable place. And — this is the part that keeps it in place — the descendants of the victims get an account in which their community was set upon by something inhuman, rather than by identifiable neighbours who were subsequently allocated their land.
I am not claiming every death in 1947 was organised. Much of it was panic, and panic is real. I am claiming that nobody has ever tried to establish the proportion, that establishing it was possible in 1948 and is nearly impossible now, and that the vocabulary everyone uses makes the question feel as though it has already been answered.
One more consequence of the missing count, and it runs through the rest of this series. Because there is no agreed number, there is no agreed proportion, and because there is no proportion, every community can hold, sincerely and without contradiction, that it suffered most. Seventy-nine years later that belief is load-bearing in the politics of three countries.
The death toll estimates run from about 200,000 to about 2 million, and the spread is a difference of method: low figures count deaths that were reported to an administration, high figures use demography and testimony to count deaths that happened.
The reporting method fails hardest exactly where the violence was worst, because the worst places produced no reporter.
No government ever attempted a count, in a state that counted cattle and ran a full census within four years. Both new countries needed a founding story that a real number would have damaged.
4What Happened to the Refugees
Two migrations, one government, and two completely different outcomes. The difference is the first entry in this series’ state ledger.
4.1 — The problem in front of them
By the end of 1947 the government of India was holding several million people who had arrived with nothing and could not go home. Setting aside for a moment how they got there, the administrative task was enormous and largely unprecedented: shelter, food, water, sanitation, disease control, and then some way of turning camps into lives.
A great deal of this was done well and it should be said plainly, because it is one of the strongest items on the credit side of the whole Congress record. Camps were established at speed. Rations moved. Epidemics that could have killed as many as the violence did not do so. A Ministry of Rehabilitation was created and given real money and real powers.
The charge in this chapter is not that the government failed to act. It is that it acted very differently in two places, and that the difference tracked something other than need.
4.2 — The exchange nobody calls an exchange
In the west, the movement was roughly two-directional and roughly simultaneous. Non-Muslims came east out of West Punjab; Muslims went west out of East Punjab and the neighbouring areas. This produced something administratively enormous and morally very strange: vacated property on both sides, in similar quantities, at the same time.
That gave the Indian state a resource. A Sikh farming family arriving from Lyallpur with a claim to land they had left behind could be allocated land left behind by a Muslim family that had gone the other way. It required a colossal paper operation — claims, verification, valuation, allotment — and India built one.
Standard acre: land measured by what it produces rather than by how big it is. Defined as the area yielding a fixed quantity of grain. Ten poor acres might count as four standard acres; four excellent acres might count as six.
It was invented for exactly this problem. A refugee’s claim was for land in a country India no longer controlled, and the land available to hand over was somewhere else with different soil and different water. Measuring both in standard acres let one office compare a claim it could not inspect with a field it could actually give away.
Why it matters here: it is a genuinely elegant piece of administration, and it worked because there was property to allocate. Where there was none, none of this machinery could run.
4.3 — And in the east, nothing to allocate
Bengal was different in three ways, and the three compound.
The movement was not simultaneous. In Punjab the exchange happened in months. In Bengal it happened over decades — a flow that continued through the 1950s and 1960s and revived at intervals afterwards. A government can plan for a wave. A trickle that never stops defeats planning, and it also defeats sympathy, because there is never a moment at which the emergency is visibly happening.
The movement was not balanced. Large numbers of Hindus left East Bengal. Muslims did not leave West Bengal in comparable numbers. So the vacated-property machinery that made Punjab resettlement possible had almost nothing to work with.
And the migrants were different. A high proportion of those leaving East Bengal in the first phase were from professional and landholding families, and later phases were overwhelmingly poor and rural, including large numbers from lower castes. The first group could resettle themselves. The second could not, and arrived after the political attention had moved on.
The outcome is not seriously disputed. Punjab’s refugees were, broadly, resettled within about a decade. Bengal’s were still being moved around the country long afterwards — to camps in central India, and to schemes far from Bengal that many of them left and walked back from. That story runs well past the boundaries of this part and it does not have a happy ending.
Two families lose everything in the same year and apply to the same government.
The first is in Punjab. Within a few years it holds an allotment: less land than before, worse land than before, but land, with a paper title. The children go to school where they now live.
The second is in Bengal. There is no allotment because there is nothing to allot. It is given a place in a camp with a dole. Ten years later it is still receiving the dole, or has been offered land eight hundred miles away in a language nobody in the family speaks.
Same catastrophe, same government, same decade. The difference is not effort or sympathy. It is that one problem had a solution lying on the ground and the other did not — and that the government did not replace the missing solution with anything of comparable weight.
That last clause is where the charge actually lives, so it is worth stating carefully before hearing the defence. Nobody is claiming the Indian state could have conjured land in West Bengal that did not exist. The claim is that a government which built an entire ministry, a claims bureaucracy and a novel unit of measurement for one problem did not build anything comparable for the other, and that the gap between the two efforts widened for twenty years while the second problem was still arriving.
Nobody disputes the outcome gap. The disagreement is about what caused it.
The instrument that solved Punjab was vacated property, and Bengal had almost none because the exchange there was one-way. Punjab’s flow also ended, allowing a finite operation to be planned and closed; Bengal’s did not, so every plan was overtaken. Add that Bengal’s inflow was still arriving when the money and the political attention had gone, and you have the whole gap without needing anybody to have wanted it.
The absence of the easy instrument explains why Punjab’s method could not be copied. It does not explain the failure to fund an alternative for two decades, nor the decision to move poor and largely lower-caste refugees to distant states rather than settle them in Bengal, nor that the western operation got a minister, a machinery and urgency while the eastern one got a dole. The constraint was real. The response to the constraint was a choice, made repeatedly.
What would settle it: the per-refugee spending figures for the two operations, year by year, which exist in budget and ministry records and have never been laid side by side in public.
Why people care so much: because it is the founding grievance of a still-live regional politics, and because the caste composition of the later eastern migration makes the neglect look like more than an accident of geography.
The reason that argument can be had at all is that one of the two operations left a great deal of paper and the other left very little, which is itself part of the story.
The evidence for the western operation is exceptionally good, for a reason that should now be familiar: it was a property transfer, and property transfers generate paper. Claim files, verification records, allotment orders and valuation registers all survive, because a state handing out land needs to know what it handed to whom.
The eastern operation generated much less, and what it generated is different in kind: camp rolls and relief payments rather than titles. A dole leaves a thinner and shorter record than a deed.
What it cannot show: the people who never entered either system. A family that crossed, was absorbed by relatives, and never filed a claim appears in no file at all. The records describe the state’s dealings with refugees, not the refugees.
4.4 — The first entry in the ledger
From here on, every part of this series carries a running record of which states were affected by the material in it and how. This chapter opens it.
| Where | What happened, 1947–1950 |
|---|---|
| East Punjab | Received several million refugees and lost a comparable Muslim population. Resettled largely through allotment of vacated land, using the standard-acre system. Lost its canal headworks relationships, its capital and much of its irrigated west. |
| West Bengal | Received a continuing inflow with almost no vacated property to allocate. Camp populations persisted for decades. Later refugees were dispersed to distant states. |
| Delhi | Population transformed within months. Large-scale killing in September 1947. Property transfers on a scale that reshaped the city permanently. |
| Assam and the east | Absorbed migration with no settlement machinery, beginning a demographic argument that is still the central fact of the state’s politics. Part Thirteen returns to it. |
| Hyderabad state | Annexed in September 1948. Mass killing followed, with no count published for sixty-four years. Chapter Six. |
| Jammu and Kashmir | Invaded, acceded, partly occupied, and frozen along a ceasefire line that still runs. Chapters Seven and Eight. |
The refugee operation in the west worked, and that belongs on the credit side. It worked because a two-way exchange left vacated property to allocate, and India built a genuinely clever machinery to allocate it.
In the east there was no vacated property, no end point, and later arrivals who were poor and largely lower-caste. The same government produced a camp and a dole instead of a settlement.
The initial gap was structural. Its survival for twenty years was a decision, taken again every year.
5Five Hundred and Sixty-Odd
On 15 August 1947 several hundred states became legally independent. Within about two years they had all gone. How that was done is the least examined success in Indian history.
5.1 — The problem, stated honestly
When the British left, paramountcy lapsed. That was not an oversight; it was written into the statute. Every princely state was legally free to join India, join Pakistan, or stand alone.
The number of them is usually given as 565. It is also given as 562, and as 584, and the disagreement is not sloppiness — it depends on whether you count tiny estates as states. Note this, because it is a small worked example of something this series meets constantly: a number everybody repeats can be a definitional choice wearing the clothes of a count.
Their sizes ranged from a few square miles to territories larger than most European countries. Some had their own currencies, armies and railways. Almost none had anything resembling representative government.
If even a modest fraction had held out, the map of India would be unrecognisable. Almost none did. That is a genuine and enormous achievement, and it belongs to Sardar Patel, who ran the States Department created in July 1947, and to V.P. Menon, the civil servant who did most of the actual negotiating.
Instrument of Accession: the short legal document by which a princely ruler joined India or Pakistan.
What it transferred was deliberately narrow: three subjects only — defence, external affairs, and communications. Everything else stayed with the ruler. Alongside it came a standstill agreement, which kept existing arrangements running while details were settled.
The narrowness was the selling point. A ruler was being asked to give up very little on paper. In practice, integration went much further within a few years, through separate agreements that merged states, ended internal autonomy and pensioned off the rulers.
Why it matters here: Kashmir’s accession is legally identical to the others. Its politics is not, and Chapter Seven is about why.
5.2 — What was promised
The rulers did not sign for nothing. They were guaranteed personal privileges, retention of private property, and a privy purse — an annual, tax-free payment from the Indian state, written into the constitution itself.
Hold that fact. In 1971 the privy purses were abolished, by a Congress government, by constitutional amendment, after an earlier attempt was struck down by the Supreme Court. Part Eight of this series deals with it. Whatever you think of hereditary payments to former autocrats — and there is a good case against them — the guarantee given in 1947 was a term of the bargain that produced the country, and it was withdrawn unilaterally twenty-four years later by the same party that gave it.
This is a forward promise and I am recording it here so it can be paid off later: Part Eight will return to this and say whether the withdrawal was justified.
5.3 — Junagadh, and the principle India chose
Junagadh was a small state on the Kathiawar coast. Its ruler was Muslim; the overwhelming majority of its people were not. In August 1947 it acceded to Pakistan, with which it shared no land border.
India rejected the accession. The argument advanced was clear and it was about people: a ruler cannot hand over a population against its wishes, and where the ruler’s religion differs from the majority’s, the majority’s wishes must govern. India moved troops to the borders, the state’s administration collapsed, the ruler left for Pakistan, and a plebiscite in February 1948 returned an overwhelming vote for India.
Now hold that principle in your hand, because Chapter Seven puts it under strain.
Three states did not resolve cleanly in 1947: Junagadh, Hyderabad and Kashmir. India’s position in each is on the record, and so is Pakistan’s.
In Junagadh and Hyderabad, both with Muslim rulers and Hindu-majority populations, India argued that the population’s wishes must govern and that a ruler could not sign away people who did not want to go. In Kashmir, with a Hindu ruler and a Muslim-majority population, India argued that the ruler’s signature was legally decisive and the accession complete — while also offering a plebiscite to confirm it.
Exactly the mirror image. In Junagadh and Hyderabad, Pakistan argued the ruler’s signature or the ruler’s choice of independence was decisive. In Kashmir it argued the population’s wishes must govern and the accession was invalid, obtained under duress while an invasion was in progress.
What would settle it: nothing, because it is not a factual dispute. Both sets of arguments are on the record and both are internally inconsistent. What can be settled is whether anyone at the time noticed the inconsistency, and the answer is that they did: it was raised in the Security Council in 1948, in those terms.
Why people care so much: because a consistent principle applied in 1947 would have given Kashmir to one country and Hyderabad to the other, and neither country can accept that trade even now.
Notice what that argument box does not contain. It contains two governments, three territories and a legal principle, and it does not contain a single person who lived in any of the three places. That absence is not an accident of how I wrote it. It is how the entire dispute has always been conducted, and it points at something underneath.
Every account of the princely states — celebratory Indian, resentful Pakistani, wistful royalist — accepts one thing without argument: that the accession was a decision made by a state.
It was a decision made by one man, in each case, who had inherited his position, ruled without a legislature worth the name, and was in most cases negotiating for his own pension and his own palace at the same time as for his people’s citizenship.
Once you see it, the whole vocabulary changes. “Hyderabad chose independence” means a Nizam chose it. “Kashmir acceded to India” means a Maharaja signed. Across several hundred states covering roughly two-fifths of the land area and tens of millions of people, the population was asked in one case, Junagadh, and only after the ruler’s decision had already been overturned by force.
Why nobody wants to notice. If accession is a ruler’s act, India’s integration is a triumph of statecraft. If accession is a population’s right, then India’s own Junagadh argument applies everywhere — including to states whose people were never asked and whose rulers signed for reasons of their own. Pakistan cannot press this either, because the same standard demolishes its case on Junagadh and Hyderabad.
So both sides argue about which ruler was coerced, and neither asks why a ruler’s signature was the operative instrument at all, in a country that had just spent thirty years arguing that people should govern themselves.
That assumption is easier to see once you know how much of this episode is documented and how little of the important part is.
The instruments of accession are published documents. So are the merger agreements that followed, and the schedules of privy purses, which had to be listed because they were payments from public funds and Parliament had to vote them.
This means the terms of the bargain are not a matter of interpretation. Anybody can read what was signed and what was guaranteed, which is why the 1971 abolition is a hard fact rather than a contested memory.
What it cannot show: what was said. V.P. Menon’s negotiations were conducted largely in person, ruler by ruler, and the assurances given in those rooms — about status, about protection, about what integration would eventually mean — survive mainly in his own later account and in the rulers’ complaints. One party to each conversation wrote the history of it.
Set against the rest of this series, this chapter is unusual: it is mostly a credit entry. A near-impossible administrative task was carried out at speed, largely without force, by two men who deserve the reputation they have. The charges here are narrow ones — an inconsistent principle, a population never consulted, and a guarantee later withdrawn — and they sit inside an achievement.
Several hundred states became legally independent in August 1947 and almost all had joined India within about two years. That was a genuine achievement, and it belongs to Patel and to V.P. Menon.
Accession transferred only defence, external affairs and communications. Rulers kept privileges, property and a constitutionally guaranteed privy purse — which a Congress government abolished in 1971. Part Eight owes you a verdict on that.
India and Pakistan each argued whichever principle delivered the territory it wanted, in all three contested cases. And in a country founded on self-rule, the population was consulted exactly once.
6Hyderabad, and the Report
A Prime Minister sent three men to find out what had happened. They found out. Then the finding disappeared for sixty-four years.
6.1 — The state that chose neither
Hyderabad was the largest of the princely states — a territory in the middle of the Indian peninsula, with its own army, currency and railway, covering land that is now spread across three Indian states. Its ruler, the Nizam, was Muslim. The great majority of his subjects were Hindu.
He chose independence. He signed a standstill agreement with India in November 1947 and played for time.
Two things were happening inside his state at once, and both belong in any honest account.
The first was the Razakars — an armed militia loyal to the Nizam’s cause, led by Qasim Razvi, which used violence against Hindus, against Communists, and against anyone advocating accession to India. Their conduct is not in serious dispute and it was ugly.
The second was a Communist-led peasant uprising in Telangana, directed against landlords and against the Nizam’s administration, which had taken over large stretches of countryside. The state was losing control of itself from two directions.
6.2 — Operation Polo
On 13 September 1948, the Indian Army entered Hyderabad. Resistance ended within five days. The Nizam surrendered and was retained as ceremonial head of state.
The operation was officially described as a police action. Part One flagged this kind of word and this is the clearest example in the series so far: a police action is what a state does inside its own territory to restore order. Calling it that settles, by vocabulary alone, the entire legal question of whether Hyderabad was a foreign state being invaded or Indian territory being brought to order. That question was genuinely contested at the time. The word decided it before the argument started, and it is still the word in most Indian textbooks.
I will use “annexation” and “military action” in this series and I am telling you that is also a choice.
Police action: a phrase for the use of armed force framed as internal law enforcement rather than as war.
The practical difference is legal, not military. War is something you conduct against another state, with laws attached. A police action is something you conduct inside your own borders, and no international law of armed conflict applies to it.
Why it matters here: five days of army operations against an entity with its own army and currency were classified, by name, as a matter of domestic order. That classification also affected what had to be reported afterwards.
6.3 — And then, what happened afterwards
The fighting ended on 18 September 1948. What followed was worse than the fighting.
Across the districts of the state, and particularly in the areas that are now in Maharashtra and Karnataka, there was large-scale killing of Muslims. It included looting, arson, mass rape, forcible conversion, and the seizure of homes and land.
The Prime Minister announced, on the Indian victory, that not a single communal incident had marred it.
He then, to his considerable credit, commissioned an inquiry.
6.4 — Three men, nine districts, three weeks
The committee was led by Pandit Sunderlal, a Congress figure of long standing, with Qazi Abdul Ghaffar and Maulana Abdulla Misri. It was appointed by Nehru. It was, deliberately, not a body of officials.
They travelled from 29 November to 21 December 1948. They went through nine of the state’s sixteen districts. They visited 21 towns and 23 large villages. They held 31 public meetings and 27 private ones, with Muslims and Hindus both.
Their conclusion, in their own framing, was a floor rather than an estimate: at a very conservative estimate, at least 27,000 to 40,000 people had lost their lives during and after the military action. In the four worst districts — Osmanabad, Gulbarga, Bidar and Nanded — they put the figure at not less than 18,000, and possibly more. In four others they put it at at least 5,000.
They also recorded that Indian troops had in places participated, and in places incited others.
This is the strongest kind of evidence available for an event of this sort, and it is worth being precise about why.
It was commissioned by the head of the government whose army was implicated, which removes the usual objection that a hostile party produced it. Its members included a senior Congress figure, which removes the objection that it came from opponents. It was based on travel through the affected districts within weeks of the events, not on later recollection. And its authors framed their own number as a conservative floor, which is the opposite of how a partisan document is written.
What it cannot show: they reached nine districts of sixteen. They were told those were the worst affected, but they were told that by the authorities, and Part One’s rule about terms of reference applies to itineraries too. Their figure covers what they could reach and verify in three weeks. It is a floor, and they said so.
6.5 — Where it went
The report was delivered. It was never published. No official explanation for that was given, then or since.
Elements of it surfaced in 1988. The full document became publicly available only in 2013 — sixty-four years after it was written, and largely through the persistence of individual scholars rather than any act of state.
Apply Part One’s rule, and then apply it again. The absence of a published death toll after Hyderabad is a decision. The suppression of a report that contained one, commissioned by the Prime Minister himself, is a second and much heavier decision, because in this case the state did not merely fail to count. It counted, wrote the number down, and then removed the number.
The range in circulation is very wide and the two ends have completely different standing. This is a live test of the grading scale.
Produced by a body appointed by the Prime Minister, on the ground within weeks, covering nine districts, and explicitly framed by its own authors as a conservative minimum. It is the only systematic contemporary attempt that exists. Grade: commission-found.
Larger estimates have circulated for decades, attributed to contemporary observers and repeated by several serious writers. They are not baseless, and they are consistent with a committee figure that covered only nine districts of sixteen and called itself a floor. But no systematic method behind them has been published. Grade: alleged.
What would settle it: district-level population and revenue records for 1948 and 1949, and the seven districts the committee never reached. Much of this material exists in state archives in Telangana, Maharashtra and Karnataka and has never been systematically examined.
Why people care so much: because the low figure makes this a bad aftermath to a justified action, and the high figure makes it the largest single massacre in the history of the Indian Union, carried out under a government whose founding claim was that it was different from what came before.
One further claim requires care. It has been written, notably by Perry Anderson and repeated by others, that Nehru suppressed the report deliberately, that Patel urged the cancellation of an ambassadorial appointment for one of its authors, and that Indira Gandhi later blocked publication as injurious to the national interest. These accounts come from serious writers. They are not, so far as I can establish, supported by a released document. Grade: Alleged. What is Solid is narrower and quite bad enough — the report existed, it was commissioned at the highest level, it was not published, no reason was ever given, and it took sixty-four years to reach the public.
Hyderabad’s ruler was Muslim, his people mostly Hindu, and his militia — the Razakars — was genuinely violent. That part of the case for acting is real and goes in.
After the army entered in September 1948, mass killing of Muslims followed. A committee appointed by Nehru himself put the toll, as a conservative floor, at 27,000 to 40,000 across nine of sixteen districts.
The report was never published and no reason was ever given. It reached the public in 2013. This is not a state that failed to count. It counted, and then removed the count.
7Kashmir: Twelve Weeks
Between the middle of October 1947 and the first days of January 1948, decisions were taken that have governed three countries ever since. Almost all of them were taken in a hurry.
7.1 — A ruler who chose neither
Jammu and Kashmir was the mirror image of Hyderabad. Its ruler, Maharaja Hari Singh, was Hindu. The majority of his subjects were Muslim. It bordered both new countries.
He did what the Nizam did: he played for time. He offered standstill agreements to both dominions. Pakistan signed one. India did not sign, and asked for talks.
Meanwhile his own state had a functioning political opposition, which is worth noting because it is often left out. The National Conference, led by Sheikh Abdullah, was a mass movement that had spent years agitating against the Maharaja’s rule. It was Muslim-led, secular in its stated politics, and hostile to the Muslim League. Its position later mattered enormously.
7.2 — October
In the third week of October 1947, armed tribesmen from Pakistan’s North-West Frontier Province crossed into the state in large numbers and advanced towards Srinagar. They looted and killed on the way, which slowed them down and is one reason they never reached the capital.
Pakistan denied involvement. Considerable evidence assembled since points to Pakistani state facilitation — transport, supplies and safe passage on a scale that raiders could not have arranged for themselves. Grade: Alleged in 1947 and treated by most specialists since as established, though Pakistan has never accepted it.
With his own forces collapsing, the Maharaja asked India for military help. India’s answer was that troops could not be sent to what was legally a foreign state. On 26 October 1947 he signed an Instrument of Accession — the same three-subject document several hundred other rulers had signed. Indian troops were airlifted into Srinagar the next day and stopped the advance.
Mountbatten, accepting the accession on India’s behalf, wrote that once law and order had been restored and the soil cleared of the invader, the question of the state’s accession should be settled by reference to the people. That sentence has been fought over ever since.
Everything downstream depends on this and both cases are serious. Here they are at full strength.
The document is legally identical in form to the instruments signed by hundreds of other rulers, all of which are accepted without question. It was signed by the recognised sovereign of the state, who under the lapse of paramountcy held exactly this power. India refused to send troops until it was signed, which is the opposite of coercion — it is a state declining to act without legal authority. And the largest popular movement in the state, the National Conference, supported accession to India.
It was signed by a hereditary autocrat with an invading army days from his capital, in circumstances where he had no alternative — the textbook definition of duress. It was signed by a Hindu ruler over a Muslim-majority population, which is the exact situation in which India itself had argued, in Junagadh weeks earlier, that the ruler’s signature does not govern. And India’s own acceptance letter said the question should be settled by reference to the people, which concedes that the signature alone was not the end of it.
What would settle it: nothing now. The plebiscite envisaged in 1948 was never held, and Chapter Eight explains why it became impossible almost immediately.
Why people care so much: because two countries have fought several wars over it, both have nuclear weapons, and each has built a national story in which its own claim is not merely stronger but obvious.
7.3 — Jammu, and the second missing count
While the fighting for the Valley was going on, something else was happening in the Jammu region to the south.
Between October and November 1947, large numbers of Muslims in Jammu province were killed or driven into West Punjab. Estimates run from about 20,000 to about 100,000 killed. In the same period, in Rajouri and in Mirpur, large numbers of Hindus and Sikhs were killed by the raiders and their allies.
Both of those events are real. Neither has ever been officially counted by anybody.
This is the second time in this part that a state which counted crops did not count bodies, and the pattern is now clear enough to state as a rule of this series: India in these years counted what it intended to administer and did not count what it intended to move past.
Take the middle of the Jammu range — say 50,000 people, which is between the published bounds and should be treated as a guess, not a figure.
Fifty thousand is a mid-sized town. It is roughly the crowd at a full cricket stadium. Killed over about six weeks, that is more than a thousand people a day, every day, for the length of a school term, in one province of one princely state, during weeks that are recorded in most histories as the background to a legal dispute about a signature.
The reason you are being given a guess rather than a number is the point of this section.
One factual question sits underneath both sides of that argument and deserves its own treatment, because a great deal of the Indian case depends on it and it is not the sort of thing that comes with a document.
The claim that Pakistan facilitated the October invasion rests on a particular kind of evidence, and it is worth seeing the shape of it. Not a confession, and not a captured order. Instead: the logistics. Several thousand armed men moved several hundred miles with fuel, transport and supplies, through territory administered by a state, along one road, without being stopped. Later scholarship has added contemporary British official reporting and the recollections of participants.
That is circumstantial, and circumstantial evidence of this type is strong for a specific reason. The question is not whether somebody helped, but whether an operation of that size could have happened without help. Very few things about 1947 are as well established as the answer.
What it cannot show: how high the decision went, or when. Facilitation by frontier officials, by parts of the army, and by the government in Karachi are three different claims, and the evidence for them is not equally strong. Pakistan has never released the relevant papers.
Twelve weeks, then. An invasion, an accession, an airlift, a war, and tens of thousands of deaths that nobody wrote down. What happened next was that India took the whole thing to New York, and Chapter Eight is about why that did not work.
Kashmir was Hyderabad in reverse: a Hindu ruler over a Muslim-majority population, playing for time, overtaken by force.
The accession of 26 October 1947 was legally the same document hundreds of rulers signed, and was signed under invasion. India’s own acceptance letter said the question should later be settled by reference to the people.
Between 20,000 and 100,000 Muslims were killed in Jammu in those same weeks, and Hindus and Sikhs were killed at Rajouri and Mirpur. Nobody has ever counted either.
8Kashmir: The Letter of 1 January
India took the dispute to the United Nations expecting an order. It received a process. The difference between those two things is the whole chapter.
8.1 — Why he went
By December 1947 the military position had stabilised and stopped. Indian troops had halted the advance and recovered ground, but could not push far before winter closed the passes. Negotiations between Nehru and Liaquat Ali Khan, with British mediation, had produced nothing.
Nehru had initially resisted taking the matter to the United Nations. He changed his mind, and on 1 January 1948 India wrote to the Security Council.
The reasoning was not stupid, and a fair account has to say so. India held a signed accession and believed it had an open-and-shut case: a member state was facilitating an armed invasion of Indian territory. Presented with that, the Council would surely order Pakistan out. The alternative was a winter of stalemate followed by a wider war with a country India was not certain it could defeat quickly, at a moment when its army had just been divided and its treasury was empty.
8.2 — What Article 35 could actually deliver
The letter was filed under Article 35 of the UN Charter, which sits in Chapter VI.
Chapter VI and Chapter VII are the two halves of the Security Council’s toolkit.
Chapter VI is “pacific settlement of disputes”. Under it the Council recommends. Its resolutions are not binding and cannot be enforced. Article 35 lets any member bring to the Council’s attention a situation whose continuance is likely to endanger international peace.
Chapter VII is “action with respect to threats to the peace”. Under it the Council can order, and can authorise sanctions and force.
Why it matters here: India filed a complaint of armed aggression into the half of the Charter that produces recommendations, not orders. Everything that followed was shaped by that choice of door.
Two consequences followed almost immediately, and both were foreseeable.
The Council treated the matter as a dispute with two parties, not a complaint with an accused. India had gone in as a victim naming an aggressor. It came out as one of two disputants. On 20 January 1948 the Council passed Resolution 39, establishing a commission to investigate — a body made up of one member chosen by India, one by Pakistan, and a third chosen by those two. That structure is not the structure of a judgment. It is the structure of a mediation, and it was set up within three weeks.
And India’s own offer became a condition. India had told the Council that although it held a valid accession, it was prepared to hold a plebiscite to confirm the people’s wishes. That was intended as evidence of good faith. It was received as a commitment, and Resolution 47 of 21 April 1948 built a sequence around it: Pakistan withdraws its irregulars, India reduces its forces, then a plebiscite under UN administration.
8.3 — Why the sequence never ran
The sequence was never completed, and the reason is structural rather than mysterious.
Step one required Pakistan to withdraw forces it had officially denied sending. Step two required India to reduce troops in territory it held, before any vote. Each side’s step one was the other side’s precondition, and neither would move first. Both had good reasons not to trust the other, and both were correct not to.
Fighting continued through 1948. A ceasefire took effect on 1 January 1949. In July 1949 a line was formally defined, with UN military observers to watch it.
That line is the single most consequential thing in this chapter. It was drawn where the armies happened to be standing when the shooting stopped. It was not drawn by reference to who lived where, or to any principle at all. It has moved only marginally in seventy-seven years and it is still, under a different name, the line today.
This is one of the most repeated charges against Nehru. It deserves a serious hearing on both sides.
India held a signed accession and a battlefield position that was improving. Taking it to the Council converted a matter India could describe as internal into a permanent international dispute with an equal party on the other side. It filed under a chapter that produces recommendations, so it got no order. It volunteered a plebiscite that became a condition binding India rather than Pakistan. And it froze the map at the moment India was gaining, ending any prospect of clearing the territory by force.
Winter had already stopped the advance; the map froze because of the passes, not the Council. Fighting on into 1948 meant a general war with Pakistan against an army India had just finished dividing, with no reserves, no money, four million refugees to feed and several hundred princely states still unresolved. The ceasefire the UN process delivered came at a time when India held the Valley — the thing it most wanted. And the plebiscite never happened, so as a binding commitment it cost India nothing except an argument it has had to have ever since.
What would settle it: the internal record of the decision — who advised what, and whether the Chapter VI point was raised before the letter went. Some of this is in published papers; the full file is not open.
Why people care so much: because it is the founding grievance of Indian foreign-policy criticism, and because it is the one moment where a single Nehru decision can plausibly be linked to seventy-seven years of consequences.
Before leaving that argument, notice what both sides of it take for granted. Both are arguing about whether a legal manoeuvre was well or badly executed. Neither asks whether the legal manoeuvre was ever the thing that decided anything.
Every participant in the Kashmir argument — Indian, Pakistani, Kashmiri, and the various outsiders — proceeds from a premise none of them defends: that the legal question is the operative one.
Watch how completely this holds. India argues the accession was validly executed. Pakistan argues it was obtained under duress and is void. Kashmiri nationalists argue neither party had the right to dispose of them. Everybody is arguing about which document, which signature, which principle. Seventy-seven years of argument about the paperwork.
But the territory has not been determined by any of that for a single day. It was determined in October 1947 by how far raiders got before Indian aircraft arrived, and then in January 1949 by where two armies were standing when they stopped. The line has held not because anyone won the legal argument but because neither side could move it at an acceptable cost.
Why nobody drops the legal frame, even though everybody can see this. Because the alternative statement — “we hold it because we hold it” — is unsayable by a democracy to its own people, and unsayable by either state to the population living there. The legal argument is not how the outcome was produced. It is how the outcome is made speakable.
This one weighs on my own book. I have spent this chapter carefully assessing an instrument of accession, and if the assumption is right, the assessment is a well-conducted analysis of something that never governed anything. I think it is worth doing anyway, because the paperwork determines what each state can say, and what a state can say constrains what it can later concede. But I want it on the page that I have written a legal chapter about a military fact.
One practical consequence of the freeze deserves recording, because it recurs. The ceasefire line cut roads, canals, grazing routes and families, exactly as the Radcliffe line had done four hundred miles to the south, and for the same reason: a line drawn by where fighting stopped has no relationship to how a place actually works. Every one of those cuts became a permanent administrative problem for somebody, and the people it fell on had no part in any of the arguments in this chapter.
India filed under Article 35, in Chapter VI — the half of the Charter that recommends rather than orders. It went in naming an aggressor and came out as one of two parties to a dispute, with a mediation commission set up in three weeks.
India’s offer of a plebiscite, made as evidence of good faith, was converted into a condition. The sequence never ran because each side’s first step was the other’s precondition.
The line was drawn where the armies were standing on 1 January 1949, by no principle at all, and it is still there. The strongest form of the charge is not that he went, but that he went badly prepared and gave away his best card on day one.
9Writing the Rules
India gave itself a constitution in January 1950. Within weeks it passed a law allowing imprisonment without trial. Both facts are in the same story and neither is a surprise.
9.1 — Who wrote it
The Constituent Assembly that produced the Indian constitution is rightly admired. It contained formidable people, it argued seriously in public, and it produced a document that has outlasted almost every constitution written anywhere in the same decade.
It was also not directly elected by the people of India, and this is left out of most accounts.
Its members were chosen by the provincial legislative assemblies. Those assemblies had been elected in 1946 on a franchise restricted by property, tax and education qualifications, which admitted somewhere around a quarter to a third of the adult population. So the body that wrote a constitution founded on universal adult suffrage was itself selected, indirectly, by an electorate that excluded most adults.
This is not a scandal and I am not presenting it as one. There was no practical alternative in 1946 and the Assembly’s own decision to adopt universal suffrage immediately, in one step, in a country with that literacy rate, was genuinely brave and was opposed by serious people. But it does bear on a claim that gets made a lot — that the constitution expressed the will of the Indian people. It expressed the judgment of a remarkable elite, acting in good faith, on their behalf.
9.2 — What was copied
Now the part that matters most for the rest of this series.
The constitution was not written on a blank sheet. Its structural base was the Government of India Act 1935 — the colonial statute that had been serving as the interim constitution. Large stretches of the machinery came across: the federal structure, the division of legislative subjects, the office of Governor, the financial provisions, the emergency architecture.
Three inheritances deserve naming individually because each becomes a weapon later in this series.
The provision for central takeover of a province. Under section 93 of the 1935 Act, a Governor could assume to himself the powers of a provincial government where administration could not be carried on in accordance with the Act. That is the direct ancestor of Article 356. It was designed by a colonial power to override elected Indian provincial ministries, and it was used for that. It was carried into the constitution of independent India substantially intact.
The emergency provisions. A structure allowing the centre to suspend the normal working of the federation. Part Nine of this series is about what happened when they were used to their limit in 1975.
Preventive detention. Not merely permitted but written into the constitution itself, at Article 22, as an express exception to the protections granted in the same article.
Preventive detention: imprisoning somebody not for something they have done but to stop them doing something in future. No charge, no trial, no conviction.
Ordinary criminal law is punitive: you are arrested for an act, tried, and convicted or released. Preventive detention runs on a prediction about the future, made by an official, and there is nothing for a court to test in the ordinary way, because no offence is alleged.
Why it matters here: most democracies allow something like it in wartime. India wrote it into its permanent peacetime constitution and used it in its first year.
9.3 — The first year
The constitution came into force on 26 January 1950. The Preventive Detention Act was passed in February 1950 — weeks later.
It was challenged almost at once. In A.K. Gopalan v. State of Madras, decided in 1950, the Supreme Court largely upheld it, reading the constitution’s protections narrowly and treating each fundamental right as a separate compartment rather than as a connected scheme. That reading stood for two decades before being overturned.
The people first detained under it included Communists — that is, political opponents.
This is not a case where one side is obviously posturing. Both have real weight.
Look at the actual year. An armed Communist insurrection was running in Telangana. Partition violence was still producing organised armed groups. A war in Kashmir had just been fought to a ceasefire. Gandhi had been assassinated by an organised conspiracy in January 1948. Several million refugees were living in camps. A state with a functioning criminal law but no capacity to run thousands of contested trials, in that year, needed a holding power, and the elected representatives of the country said so openly and voted for it.
Every emergency argument is made in an emergency; that is what the word means. The test is what happens when the emergency ends, and this power never ended — it has been renewed, replaced and re-enacted in different forms continuously ever since, through decades containing no insurrection. And these were people who had themselves been detained without trial by the British, who knew exactly what the power felt like from the inside, and who wrote it into the permanent constitution rather than into a temporary statute with an expiry date.
What would settle it: the Constituent Assembly debates, which are published in full, record who argued for a time limit and who resisted one. Anyone can read them. Almost nobody has.
Why people care so much: because every Indian government since has used this power, and because the party that wrote it into the constitution is the party that, twenty-five years later, used its descendants to jail its opponents wholesale.
9.4 — Article 370, and the promise attached to it
One more provision belongs here because Chapter Eight left it hanging.
Article 370 gave Jammu and Kashmir a special constitutional position: its own constitution, and a limit on Parliament’s power to legislate for it without the state’s concurrence. It was placed in the part of the constitution headed “Temporary, Transitional and Special Provisions”. The word temporary is in the document.
Two things follow and both are load-bearing for later parts. First, the arrangement was presented as provisional pending a final settlement — which is consistent with the plebiscite language of 1947 and 1948 and inconsistent with treating the accession as having closed the question. Second, it was hollowed out gradually over the following decades by a long series of orders and amendments, most of them made by Congress governments, until relatively little of the original autonomy remained well before it was formally removed in 2019.
Recording that as a forward promise: Part Seven of this series traces the hollowing-out year by year, and Part Eighteen deals with the removal.
Everyone who argues about the Indian constitution — those who admire it and those who think it too centralising — shares a premise: that it was designed. That a group of people sat down, decided what kind of state they wanted, and built it.
Much of it was not designed. It was edited. The base text was a colonial statute, and the parts that survived the edit least changed are precisely the parts concerned with control: the takeover of provinces, the emergency architecture, detention without trial.
Here is why that is more than a technical point. The freedom movement told itself for thirty years that the problem was the British. Not the district magistrate’s powers — the fact that a foreigner held them. Not the detention law — the fact that it was used against nationalists. If the problem is who holds the instrument, then independence solves it completely, and keeping the instrument is not merely acceptable but sensible, because now it is in the right hands.
If the problem was the instrument, independence solved nothing about it, and the freedom movement’s own framing had made that conclusion nearly unthinkable to the people best placed to reach it.
The test is what happened next, and it runs the length of this series. The powers were kept. The powers were used. Article 356 was used against elected state governments dozens of times. Detention without trial never lapsed. In 1975 the emergency architecture was used to suspend the constitution itself, by the party that wrote it.
Both sides of the ordinary argument miss this in the same way. The admirer credits the framers with designing a durable democracy. The critic blames them for designing an over-mighty centre. Neither notices that on the specific question of coercive power, the framers largely did not design at all. They inherited, and left it running.
That is a heavy charge to lay at the door of people who cannot answer it, so it is worth being clear about how much of their thinking is actually recoverable.
For this chapter the evidence is close to ideal, and almost nobody uses it. The Constituent Assembly debates were recorded verbatim and are published in full. You can read who proposed preventive detention, who objected, what the objections were, and how the vote went. The same is true of the parliamentary debates on the Preventive Detention Act weeks later.
There is no need to infer intent here. The framers said what they thought, at length, on the record, knowing it would be published.
What it cannot show: what was settled outside the chamber. A great deal of the constitution was worked out in committee and in private negotiation, and the floor debate is sometimes the ratification of a decision already taken rather than the making of one. Where the debate is thin on a major provision, that is often the sign.
9.5 — Promises entered into the record
Part One committed this series to naming forward promises where they are made, so that later parts can be checked against them. Four were made in the period covered by this part, and all four come due later.
The privy purses and the guarantees to the rulers, written into the constitution as the price of accession. Withdrawn in 1971. Part Eight owes a verdict.
The plebiscite in Kashmir, offered by India to the Security Council in January 1948 and never held. Part Seven owes an account of when and why the offer stopped being live.
Article 370 as a temporary provision. Temporary is the constitution’s own word. Part Seven owes the year-by-year record of what happened to it, and Part Eighteen owes the 2019 removal.
Detention without trial as an emergency measure. Justified in 1950 by circumstances everybody could see. Part Nine owes the account of what it became in 1975.
Write these down. A series that makes promises and does not pay them is doing the thing it accuses others of.
The Constituent Assembly was indirectly chosen by assemblies elected on a restricted franchise. It then granted universal adult suffrage in one step, which was brave. But it did not express the will of the people; it acted on their behalf.
The constitution’s base text was the Government of India Act 1935, and what survived least changed was the machinery of control: central takeover of provinces, the emergency provisions, and preventive detention written into the constitution itself.
The Preventive Detention Act was passed weeks after the constitution came into force, and was used against political opponents in year one. The emergency justified a power. It did not justify making it permanent, and that part cost nothing to get right.
10An Honest List of What We Do Not Know
This part contains more missing numbers than any other in the series. That is not a weakness of the research. It is the finding.
10.1 — Genuinely unknown
Five things in this part cannot be established, and in four of the five cases the reason they cannot be established is more interesting than the gap.
How many people died in the Partition violence. The published range runs from about 200,000 to about 2 million. The low end counts what reached an administration; the high end reasons from demography and testimony.
Why it is unknown: nobody tried. Not in 1948 with the camps full, not in the 1951 census, not since. Two new states each needed a founding account that a real number would have damaged.
How many were killed in Jammu in October and November 1947. Published estimates for Muslims killed run from about 20,000 to about 100,000, a factor of five. Killings of Hindus and Sikhs at Rajouri and Mirpur in the same weeks have no published figure at all.
Why it is unknown: the territory was a war zone under a princely administration that was collapsing, and afterwards it belonged to a dispute that both governments wanted to argue in terms of law rather than blood.
How many were killed after the Hyderabad action, above the committee’s floor. The Sunderlal committee reached nine districts of sixteen and described its own 27,000 to 40,000 as a conservative minimum. What the other seven districts contained has never been established.
Why it is unknown: the committee had three weeks and was directed to the districts the authorities named. Then the report was suppressed, so no follow-up was ever commissioned.
Why the Sunderlal report was suppressed. No document explaining the decision has been released. Serious writers have attributed it to Nehru, to Patel’s urging, and later to Indira Gandhi’s refusal to publish. Grade: Alleged. I am not going to convert that into a finding by repeating it confidently.
Why it is unknown: because the papers have not been released, which is itself the second-order version of the same decision.
What proportion of the 1947 violence was organised. The evidence that some of it was organised is strong: coordinated timing, ambushes at chosen points, armed bands moving between districts, participation in places by men in uniform. The proportion is unknown.
Why it is unknown: establishing it required contemporaneous investigation, and there was none. This one is now close to unrecoverable, and it is the single most consequential thing this part cannot tell you.
10.2 — Solid
The documentary chronology. The award finished on 12 August, submitted on the 13th, circulated on the 16th, published on the 17th. The accession signed on 26 October 1947, the airlift on the 27th. India’s letter to the Security Council on 1 January 1948. Resolution 39 on 20 January, Resolution 47 on 21 April. Ceasefire on 1 January 1949. The constitution in force on 26 January 1950. None of this is in dispute anywhere.
The legal texts. What the Instrument of Accession transferred, and that Kashmir’s was in the same form as the others. That Article 22 authorises preventive detention. That Article 356 descends from section 93 of the Government of India Act 1935. That Article 370 sits under a heading containing the word “temporary”. These are readable by anybody and they settle several arguments that are still conducted as though they were open.
That India filed under Article 35, in Chapter VI. A matter of record, and it determined that the Council could recommend and not order.
The Sunderlal committee’s existence, appointment and finding. Commissioned by Nehru. Three named members. On the ground from 29 November to 21 December 1948, across nine districts, 21 towns and 23 large villages. Conservative floor of 27,000 to 40,000 dead. Not published until 2013. Grade: Commission-found, and it is the strongest single document in this part.
The integration of the princely states. Substantially complete within about two years, achieved mainly by negotiation. This is solid and it is a credit item.
The asymmetry between the two refugee operations. Not disputed by anybody, including by those who explain it as structural rather than chosen.
Who wrote the constitution. The Assembly’s composition, and the franchise of the 1946 elections that produced the bodies which chose it, are matters of record.
10.3 — The three files
Most of section 10.1 could be closed by three sets of papers that exist and are not open.
The full record of the meeting of 16 August 1947, at which the boundary award was put in front of the Indian and Pakistani leaderships. It would establish whether anybody in that room argued for immediate publication.
The Sunderlal committee’s underlying material and the decision not to publish. Sixty-four years of suppression is documented. The reasoning is not.
The internal advice preceding the letter of 1 January 1948. Whether the distinction between Chapter VI and Chapter VII was put to Nehru before the letter went, and by whom, is the difference between an error and an oversight.
None of these is a national security matter in 2026. All three are older than most of the people they would embarrass, and every one of those people is dead.
Genuinely unknown: the Partition toll, the Jammu toll, the Hyderabad toll above the committee’s floor, why the report was suppressed, and how much of the violence was organised.
Solid: the documentary chronology, the legal texts, the Chapter VI filing, the Sunderlal committee’s finding, the integration of the states, the refugee asymmetry, and the composition of the Constituent Assembly.
Three closed files would settle most of the first list. None of them is a security matter now, and everybody they could embarrass is dead.
Sources & further reading — Part 2
Timeline, 1946–1950
Dates only. The arguments are in the chapters.
| Date | What happened |
|---|---|
| Aug 1946 | Killing in Calcutta leaves roughly four thousand dead and about a hundred thousand homeless. The violence of Partition begins before the decision to partition. |
| Mar 1947 | Mountbatten arrives with a mandate to end British rule by June 1948. |
| 3 Jun 1947 | The partition plan is announced. Independence is brought forward to 15 August 1947. |
| 8 Jul 1947 | Radcliffe reaches Delhi. He has about five weeks and has never been to India. |
| Jul 1947 | The States Department is created under Patel, with V.P. Menon as secretary. The Punjab Boundary Force deploys, around 55,000 strong. |
| 18 Jul 1947 | The Indian Independence Act is passed. Paramountcy will lapse; princely states become legally free. |
| 12–13 Aug | Radcliffe completes the award and submits it sealed to the Viceroy. |
| 14–15 Aug | Pakistan and India become independent dominions. The border is not published. |
| 16 Aug | The award is shown to both leaderships. Both object. Both agree to let it stand. |
| 17 Aug 1947 | The award is published. People near the border learn which country they are in. |
| Aug–Nov 1947 | Mass killing and migration across Punjab and Bengal, and in Delhi. Between 200,000 and 2 million dead; 12 to 20 million displaced. No count is attempted. |
| Sept 1947 | Junagadh’s accession to Pakistan is rejected by India on the principle that the population’s wishes govern. |
| Oct–Nov 1947 | Killings in Jammu province: between 20,000 and 100,000 Muslims. Hindus and Sikhs killed at Rajouri and Mirpur. Neither counted. |
| 22 Oct 1947 | Armed tribesmen from Pakistan’s frontier enter Jammu and Kashmir and advance on Srinagar. |
| 26 Oct 1947 | The Maharaja signs the Instrument of Accession. India airlifts troops to Srinagar the next day. |
| Nov 1947 | India signs a standstill agreement with Hyderabad. The Nizam plays for time. |
| 1 Jan 1948 | India writes to the UN Security Council under Article 35, in Chapter VI. |
| 20 Jan 1948 | Resolution 39 sets up a commission — a mediation body, not an enforcement one. |
| 30 Jan 1948 | Gandhi is assassinated by an organised conspiracy. |
| Feb 1948 | A plebiscite in Junagadh returns an overwhelming vote for India. It is the only time any princely population is asked. |
| 21 Apr 1948 | Resolution 47 sets a sequence: Pakistan withdraws, India reduces, then a plebiscite. It is never completed. |
| 13–18 Sep 1948 | Operation Polo. The Indian Army enters Hyderabad; resistance ends in five days. |
| Sept–Oct 1948 | Mass killing of Muslims across Hyderabad state. Nehru states publicly that no communal incident marred the operation. |
| 29 Nov–21 Dec 1948 | The Sunderlal committee travels through nine of sixteen districts. |
| 1 Jan 1949 | Ceasefire in Kashmir. The line stands where the armies stopped. |
| 1949 | The Sunderlal report is delivered. It is not published. It becomes public in 2013. |
| Jul 1949 | The ceasefire line is formally defined, with UN observers to watch it. |
| 26 Nov 1949 | The Constituent Assembly adopts the constitution. |
| 26 Jan 1950 | The constitution comes into force. India becomes a republic. |
| Feb 1950 | The Preventive Detention Act is passed, weeks later. Communists are among the first detained. |
Part Three starts here. It opens in 1951, with a government holding a constitution fifteen months old and preparing to amend its guarantee of free speech.
Glossary
Every hard word used in this part, in plain English.
| Word | What it means |
|---|---|
| Article 22 | The part of the Indian constitution which, alongside protections for arrested people, expressly permits preventive detention. |
| Article 356 | The provision allowing the centre to dismiss an elected state government. Descended from section 93 of the Government of India Act 1935. |
| Article 370 | The provision giving Jammu and Kashmir a special constitutional position, placed under a heading including the word “temporary”. |
| Boundary award | The published decision of a boundary commission fixing a border. Borrowed from arbitration: the parties agree in advance to accept it. |
| Chapter VI | The half of the UN Charter dealing with peaceful settlement. The Security Council can recommend, not order. |
| Chapter VII | The half dealing with threats to peace. Here the Council can order, and authorise sanctions or force. |
| Communal | In Indian usage, relating to conflict between religious communities. Names the participants and says nothing about organisation. |
| Dominion | A self-governing country within the Commonwealth with the British monarch as formal head of state. India was one until January 1950. |
| Evacuee property | Land and buildings left behind by people who migrated at Partition, taken over by the state and used to compensate incoming refugees. |
| Instrument of Accession | The document by which a princely ruler joined India or Pakistan, transferring three subjects only: defence, external affairs and communications. |
| Paramountcy | British control of the princely states’ defence and foreign affairs by treaty. It lapsed in 1947 rather than transferring. |
| Plebiscite | A direct vote of a population on a single question, such as which country to join. |
| Police action | Armed force framed as internal law enforcement rather than war. The classification is legal, not military. |
| Preventive detention | Imprisonment to stop somebody doing something in future. No charge, no trial, no conviction. |
| Privy purse | The annual tax-free payment guaranteed to former princely rulers, written into the constitution and abolished in 1971. |
| Razakars | The armed militia loyal to the Nizam of Hyderabad, which used violence against Hindus, Communists and supporters of accession to India. |
| Standard acre | Land measured by what it yields rather than its area, used to match refugee claims against available land. |
| Standstill agreement | A holding arrangement keeping existing administrative ties running while accession terms were settled. |
What Comes Next
Part Three — The Constitution Amended, 1950 to 1959
Part Three takes the first decade of the republic. It asks:
- Why did a government amend its own brand-new guarantee of free speech within fifteen months of adopting it, and who was it aimed at?
- What was the Ninth Schedule created to protect, what ended up inside it over the following decades, and was the drift foreseeable in 1951?
- The first general election was the largest ever held anywhere on Earth. How was it actually run, what did it cost, and what did the government learn about its own reach?
- Land reform was legislated across most of India in this decade. What happened between the statute and the field, and which states were the exceptions?
- On what grounds was the elected government of Kerala dismissed in 1959, and did those grounds meet the test written into Article 356?
- By 1959, who inside Congress was still able to tell the leadership it was wrong, and what had happened to the ones who tried?
Questions, not topics. If any of them is answered badly, you will be able to tell.