The Cut

1The arithmetic

The first question this whole work was asked to answer is: why was Punjab divided? The answer is not a conspiracy or a betrayal. It is a census table, and the census table was nine hundred years in the making.

The numbers

By the 1941 census, undivided Punjab was approximately:

Now recall Part 3, Chapter 5, which explained where that Muslim majority came from and — crucially — where it was.

Punjab was not a province with a uniform 53% Muslim population. It was a heavily Muslim west — Multan, Rawalpindi, Jhelum, the Lahore division, many districts over 80% Muslim — joined administratively to a Hindu and Sikh east: the Jalandhar Doab, Ambala, Ludhiana, the Malwa country south of the Sutlej.

That division was formed over nine hundred years, by the mechanisms set out in Part 3: the west came under Muslim rule from around 1030, 160 years before the east; the great Sufi centres were in the west; and Brahminical institutional infrastructure was denser in the east.

Why the division followed mechanically

Once the principle was accepted — first in the Muslim League's Lahore Resolution of 1940, then definitively in the 3 June Plan of 1947 — that British India would be partitioned according to religious majority, Punjab's fate was arithmetic rather than choice.

• A province with a Muslim majority could not go wholly to India.
• A province with a 44% non-Muslim minority, concentrated in a contiguous eastern block, could not go wholly to Pakistan.

So Punjab was divided. Bengal, for exactly the same reason, was divided too. No other province was, because no other province had this configuration.

And the part that made it a catastrophe for one community

Now apply that arithmetic to the Sikhs, and you find the specific tragedy of 1947 in Punjab.

• Sikhs were about 15% of Punjab and roughly 2% of India.
• They were not a majority in a single district anywhere in the province. Not one. Their population was spread across central Punjab in a band, dense in places, dominant nowhere.
• Their sacred geography was split down the middle. Amritsar and the Golden Temple in the east. Nankana Sahib — Guru Nanak's birthplace — in the west. Panja Sahib in the far north-west. Kartarpur, where Nanak died, in the west. Dera Sahib in Lahore, where Guru Arjan was martyred.
• Their most valuable landholdings were in the canal colonies of the west.

Put those four facts together and you get a conclusion that ought to be stated as flatly as possible:

That is why the Sikh leadership's position in 1946 and 1947 looks, in retrospect, like a man being asked to choose which of his own limbs to keep. Chapter 2 is about what they actually chose, and why.

2The Sikh position, and the word "Khalistan"

Four options were on the table between 1940 and 1947. Understanding why three of them failed is essential — and one fact in this chapter reshapes the entire argument of Part 11.

Option 1: an undivided India

This was the Akali Dal's actual preference and it was consistent. Master Tara Singh, the pre-eminent Sikh political leader of the period, opposed the creation of Pakistan and campaigned against it. On 3 March 1947, outside the Punjab Assembly in Lahore, he publicly raised the slogan against Pakistan before a hostile crowd.

The option failed because it was not in Sikh hands. Once Congress accepted partition in principle, it was over.

Option 2: join Pakistan

This was a real option, seriously offered. Jinnah made approaches to Sikh leaders with proposals of autonomy and constitutional safeguards. Some Sikh figures considered it.

It was rejected, for three reasons that were all rational at the time:

• Historical memory. Part 4 is not distant history to Sikhs; it is recited daily. A state defined by Islam did not look like a safe home to a community whose founding narrative includes two Gurus executed by a Muslim empire and four children killed at Sirhind. Whether that inference was fair to Jinnah's Pakistan is a separate question. It was the inference that was drawn.
• The Muslim League's recent record in Punjab. The communal campaigning of 1946 and the violence of early 1947 had destroyed whatever trust existed.
• The arithmetic. Sikhs would have been roughly 2% of Pakistan, with no leverage and constitutional guarantees enforceable by nobody.

Option 3: Azad Punjab

This is the proposal the Akali Dal actually put forward, from around 1943, and it deserves far more attention than it gets because it was constitutionally interesting.

Azad Punjab — "Free Punjab" — proposed redrawing the province's boundaries by detaching some heavily Muslim western districts and some heavily Hindu eastern ones, so that the resulting province would have no community with an absolute majority.

The logic was elegant. If nobody has 51%, everybody must negotiate. A Sikh bloc of perhaps 20% becomes a permanent, decisive balancing force — never able to rule, never able to be ignored. Power-sharing is not a matter of goodwill; it is arithmetically compulsory.

Note what this was: a proposal for a differently-drawn province inside India. Not separation. Not a Sikh state. A constitutional mechanism to make domination impossible.

It was rejected by both Congress and the Muslim League, and it died.

Option 4: a separate Sikh state — and where the word came from

Now the fact that changes things.

The word "Khalistan" was first used as the title of a pamphlet published in 1940 by V. S. Bhatti, a doctor from Ludhiana, shortly after the Muslim League's Lahore Resolution. That appears to be its first recorded appearance anywhere.

And here is what happened to it:

Master Tara Singh denounced it. The pre-eminent Sikh political leader of the era publicly rejected the pamphlet, on the grounds that it added confusion to the confusion the Muslim League had already created.

Why a Sikh state was impossible anyway

Even setting aside what the leadership wanted, the option was not available, for reasons that are arithmetic rather than political:

• No Sikh-majority district existed. A state has to be somewhere, and there was nowhere with a Sikh majority to put it.
• No contiguous territory. The Sikh population ran in a band through central Punjab, mixed everywhere with Hindus and Muslims.
• No sponsor. Neither the British, nor Congress, nor the League had any interest in creating a third dominion.
• And the sacred sites were on both sides of any possible line — so even a successful Sikh state would have lost Nankana Sahib.

March 1947: the killing begins before independence

In March 1947, months before either country existed and five months before Radcliffe's line was published, organised massacres of Sikhs and Hindus swept the Rawalpindi division — Rawalpindi, Attock, Jhelum — and parts of Multan. Villages were attacked systematically. Thousands were killed.

At Thoa Khalsa, in Rawalpindi district, around ninety Sikh women died in a well rather than be taken. I state it because it happened and because it is remembered; I am not going to elaborate on it.

The significance of March 1947 for the argument of this part is this: the violence did not begin when the border was announced. It was already running, in an organised form, five months earlier. Chapter 4 returns to this.

3The Radcliffe Line

A London barrister who had never been to India was given five weeks, outdated maps and no site visits, and told to decide where fourteen million people would live. The result was published two days after independence.

The facts of the process

The two days that killed people

That last row deserves to be sat with.

India became independent on 15 August 1947. Pakistan on 14 August. The boundary between them was not published until 17 August.

So for the first days of both countries' existence, millions of people in Punjab did not know which country they were in. Villages celebrated independence without knowing whose independence it was. Families made decisions about whether to move based on rumour. Administrators did not know whose orders to follow. Police did not know who they worked for.

The delay is generally attributed to Mountbatten's desire not to have the announcement mar the independence celebrations. Whatever the reason, the effect was that the most dangerous information in the subcontinent was withheld from the people it concerned during precisely the days when the violence was accelerating.

The two controversies

Gurdaspur. Gurdaspur district had a narrow Muslim majority overall, and most of it was awarded to India. Pakistani critics have long argued this was deliberate, to give India a land route to Kashmir — which it did.

The case for the accusation: the district had a Muslim plurality; the award gave India the only viable road access to Jammu and Kashmir; and Mountbatten had both motive and access to Radcliffe.

The case against: at tehsil level — the unit Radcliffe actually worked with — the picture is more mixed, with some tehsils non-Muslim majority. The Madhopur headworks on the Ravi, which fed the canal system of eastern Punjab, were also in play. And Radcliffe's terms of reference explicitly allowed "other factors."

Ferozepur. The Ferozepur headworks on the Sutlej were awarded to India, giving India control of water feeding canals in West Punjab. There is evidence in the record that the line here was altered late in the process.

I cannot resolve either dispute and I am not going to pretend to. Both remain genuinely contested by serious historians, and the documentary record is incomplete because Radcliffe burned his working papers before leaving India. He also declined his fee.

What Radcliffe actually got wrong

Here is my assessment, and it is more specific than "he drew a bad line."

The general placement was largely unavoidable. The line ran roughly down the middle of the Bari Doab, between Lahore and Amritsar, because that is approximately where the Muslim-majority west met the Hindu-Sikh-majority east. As Part 3 showed, that fault line was nine hundred years old. Radcliffe did not invent it. He traced it.

The failures were procedural, and each one had a body count:

1. Five weeks for a task requiring years.
2. No site visits, so no knowledge of which villages actually shared wells, roads and irrigation channels.
3. Six-year-old census data as the sole demographic basis.
4. Publication two days after independence, which left millions in the dark during the most dangerous week.
5. No population transfer plan and no adequate security arrangement, which is Chapter 4.

4The killing

This chapter gives the numbers, explains why they cannot be made precise, describes the mechanics, and then addresses the question everybody asks and nobody should.

The numbers, with their ranges

Why the death toll cannot be established

A factor-of-ten range is unusual even for historical atrocity figures, so the reasons are worth stating:

• There was no functioning administration. The civil service, police and courts were themselves being partitioned, and their staff were themselves fleeing.
• The victims were in motion. People died on trains, on roads, in fields, in canals, hundreds of kilometres from anywhere they were registered.
• Nobody was counting. No systematic effort at enumeration was made at the time by either new state.
• And both new governments had reasons not to look too hard. A precise figure would have been an indictment of the transfer of power that both had just accepted, and of their own failure to protect their citizens.

What is not in dispute is the order of magnitude: hundreds of thousands dead, and the largest forced migration in recorded human history, most of it inside one province, in roughly three months.

The mechanics: what made it so lethal

Trains. Refugee trains ran in both directions and were attacked in both directions. There are multiple documented cases of trains arriving at their destination with everyone aboard dead.

The kafilas. Foot columns of refugees — sometimes tens of thousands of people moving together, with carts and cattle, at walking pace, for weeks. They were attacked on the road, and they were almost impossible to defend.

Organised formations, not just mobs. This is the point I most want to establish, because it is routinely softened. The violence was substantially carried out by organised armed groups with command structures, on all three sides. The Muslim League National Guard. Hindu organisations including the RSS. Sikh jathas, some organised around ex-servicemen. These groups planned attacks, coordinated across villages, and in many cases had lists.

Demobilised soldiers. Part 6's recruitment figures become a weapon here. Punjab was full of men who had recently returned from the Second World War with infantry training, and in many cases with access to weapons.

And no authority. The Punjab Boundary Force, about 25,000 men under Major-General T. W. Rees, was supposed to police twelve districts containing some fourteen million people. It was a mixed-community force, and it fractured — many of its soldiers had families among the victims on one side or the other. It was overwhelmed almost immediately and was disbanded on 1 September 1947, about a month after it began.

"Who started it?"

Now the question, and my answer to it.

The honest response is that it cannot be established, and that in practice the question is almost always asked as a way of avoiding something.

Here is what the record actually shows:

• Organised massacres of Sikhs and Hindus in the Rawalpindi division took place in March 1947 — five months before independence, before Radcliffe's line existed.
• Organised massacres of Muslims in East Punjab — Jalandhar, Ludhiana, Amritsar, Gurdaspur, Ambala — in August and September 1947 were as systematic, as planned and as lethal.
• Communal killing in Punjab had precedent going back through the 1920s and 1930s, and both Bengal and Bihar had already seen mass violence in 1946.

Anyone can find an earlier atrocity to justify a later one. That is what a cycle is. Setting the clock's start point is a choice, and everybody chooses the point that makes their own community the responder rather than the initiator.

Malerkotla

One exception is worth recording, because it shows what was possible.

Malerkotla was a Muslim-ruled princely state in East Punjab. During Partition, it saw essentially none of the massacre that engulfed the districts around it, and its Muslim population largely remained. It is today the one significant Muslim-majority town in Indian Punjab.

The explanation given in Punjabi tradition goes back to Part 4. When Guru Gobind Singh's two youngest sons were sentenced at Sirhind in 1704, Sher Mohammad Khan, the Nawab of Malerkotla, raised a formal protest against the execution of children — the haa da naara, the cry for justice. The tradition holds that the Guru blessed his house for it.

Whether the causation is that simple, historians can argue. What is not in doubt is that in 1947, when almost everywhere else in East Punjab was killing, Malerkotla did not, and Sikhs in the surrounding area actively protected it.

5The ledger

Every part of India lost something in 1947. Punjab lost a category of things that no other region lost, and it is worth listing them explicitly, in both directions.

What went west, to Pakistan

The scale of the religious loss in particular has no parallel. The majority of the most sacred sites of Sikhism ended up in another country, one that Indian Sikhs would need a visa to enter and which has been at war with India four times since.

What went east, to India

The other side of the ledger is equally real and gets far less attention in Indian writing, so it goes here explicitly.

Around eight million Punjabi Muslims lost their homes, land, businesses and ancestral villages in Jalandhar, Ludhiana, Ambala, Hoshiarpur, Gurdaspur, Amritsar and Ferozepur. Families that had farmed the Doab for centuries walked west with what they could carry. The shrines, mosques and graves they left behind are in India.

Their descendants are Pakistani Punjabis today, and a great many of them can still name the village in Indian Punjab their grandparents came from.

And what happened to the language

One loss is rarely counted and it is substantial. Punjabi was split into two scripts and two states.

• In India: written in Gurmukhi, and eventually the official language of a Punjabi-speaking state — after the struggle described in Part 8.
• In Pakistan: written in Shahmukhi, a Perso-Arabic script. And Pakistani Punjab adopted Urdu as its official language.

The result is that in the province containing the largest number of Punjabi speakers on earth, Punjabi has no official status and is not the medium of education. And a Punjabi in Amritsar cannot read a book printed in Lahore, or vice versa, without learning a second script — despite speaking the same language.

6The water, within eight months

Part 1 predicted this in the Bronze Age. Part 6 built the machine. Part 7 cuts it in half — and the first crisis arrives before Punjab has finished burying its dead.

The problem, immediately

Part 6, Chapter 3 described the canal system the British built: an integrated network designed as a single machine, with headworks controlling flow to canals hundreds of kilometres downstream.

The Radcliffe Line cut that machine in half — and, critically, it left the headworks in India and much of the irrigated land in Pakistan:

• Madhopur headworks on the Ravi fed the Upper Bari Doab Canal, which watered both Amritsar (India) and Lahore (Pakistan).
• Ferozepur headworks on the Sutlej fed the Dipalpur Canal into West Punjab and Bahawalpur, and the Bikaner Canal into India.

What happened

20 December 1947: the chief engineers of East Punjab and West Punjab signed a Standstill Agreement — the existing distribution of water would continue until 31 March 1948, covering the rabi crop.

1 April 1948: the day after the agreement expired, East Punjab stopped the flow. Canals in West Punjab — including the Central Bari Doab Canal serving Lahore — ran dry.

18 April 1948: an agreement was reached between the two sides.

30 April 1948: water was restored.

A month, in the sowing season, in a semi-arid region. The economic damage and the political damage were both severe, and the episode established for Pakistan a permanent strategic anxiety about upstream control that has never gone away.

What it led to

The 1948 crisis began the negotiations that produced the Indus Waters Treaty of 1960, brokered by the World Bank. Its allocation:

7What 1947 did to Punjabi politics permanently

Five structural consequences, all of which are still operating. The first one surprised me most.

1. It created a Sikh-majority Punjab for the first time in history

Before 1947, Sikhs were about 15% of undivided Punjab and were not a majority in a single district.

After the population exchange, with roughly six million Hindus and Sikhs arriving in East Punjab and eight million Muslims leaving, the Sikh share of the remaining province rose dramatically. After the further reorganisation of 1966 — Part 8 — Sikhs became a majority in the state of Punjab, at roughly 57–58% today.

I find this genuinely important and rarely stated. The political fact that underpins every argument in Parts 8 through 12 — that Punjab is a Sikh-majority state — is younger than most of the people arguing about it, and it was created by the two events that Punjabis remember as catastrophes.

2. It destroyed the Unionist Party

From 1923 to 1947, Punjab was governed largely by the Unionist Party — a cross-communal alliance of Muslim, Hindu and Sikh landholders, organised around agrarian interest rather than religion. Under Sikandar Hayat Khan it was the dominant force in the province and had specifically kept the Muslim League weak in Punjab.

It collapsed in the 1946 elections as communal polarisation overwhelmed class interest. Its destruction is a precondition of partition, not just a casualty of it: as long as Punjab had a functioning cross-communal governing party, the province could not be mobilised along purely religious lines.

Nothing has replaced it. No cross-communal governing formation has held Punjab since.

3. It created a refugee politics

East Punjab absorbed millions of people who had lost everything, arrived with nothing, and carried a verified first-hand grievance. That energy — hard, mobile, unsentimental, willing to take risks because there was nothing left to lose — went into Punjabi politics, business and agriculture for decades. The Green Revolution's success in Punjab owes a great deal to refugee farmers rebuilding from zero.

4. It established the precedent: when Punjab has a problem, Punjab gets smaller

This is the one with the longest reach.

Twice in nineteen years, Punjab's political problems were resolved by redrawing its boundaries along community lines. Twice, the outcome was a smaller Punjab.

That precedent matters enormously for Parts 11 and 12, and it cuts in two directions at once. It is the reason a section of Punjabi opinion regards further territorial argument as the natural way to seek redress. It is also the reason another section regards it with dread, having watched where the last two rounds led.

5. It gave everyone a verified grievance

After 1947, nobody in Punjab needed to be persuaded that mass atrocity was possible, that the state could fail to protect them, or that their neighbours could turn. They had seen it, personally, within living memory.

That is the emotional ground on which Parts 9 and 10 take place. When people in 1984 reacted to events with a speed and an absoluteness that outsiders found baffling, it was not because they were irrational. It was because a very large number of them, or their parents, had already lived through the thing they were afraid of.

8Scorecard: what Part 7 proves

Timeline of Part 7

What happens in Part 8

Part 8, Redrawing Punjab, covers 1947 to 1980, and it is the part where the modern crisis is actually built. Everything people argue about when they argue about 1984 was decided in these thirty-three years.

It covers the Punjabi Suba movement — a fifteen-year campaign for a Punjabi-speaking state, which raises the uncomfortable question of why Punjabi Hindus declared Hindi as their mother tongue in the 1951 and 1961 censuses. It covers the 1966 reorganisation that created Haryana and Himachal and reduced Punjab to a seventh of its historical size. It covers Chandigarh — a capital built for a province that no longer exists, promised to Punjab repeatedly and never delivered. It covers the river waters and the origins of the Sutlej-Yamuna Link canal.

It covers the Green Revolution, which made Punjab rich and India food-secure, and which planted every ecological and economic problem that Part 13 will have to account for.

And it covers the Anandpur Sahib Resolution of 1973 — the document at the centre of everything that follows, which is quoted constantly by people who have not read it, and which I will reproduce and examine clause by clause.

Above all, Part 8 applies the broken-bargain test from Part 6 to the Indian state's dealings with Punjab. What was promised — on language, territory, the capital, the rivers and constitutional status? What was delivered? And how much of what happened afterwards is explained by the gap?