The Part That Made Me Angry
I promised at the start of this series that I would not let my own feelings drive the writing. This is the part where that promise cost me the most.
Part 2 was about hymns and archaeology. You can argue about a metaphor for a long time without anyone getting hurt. Part 3 is different. Part 3 is where we reach texts that specify, in plain language, which body part to mutilate and for which offence.
I sat with those verses for a while. Then I did the thing this series requires and went looking for the strongest case that they were not what they appear to be. And I found real arguments — better than I expected. Some of them I now think are correct.
So this part has an unusual structure. Chapter 3 prints what the text says — the worst of it, in full, without softening. Chapter 4 then puts the best possible case that it was never the law of the land. I have kept those two chapters separate on purpose, because the most common failure in this argument is to do one and skip the other.
If you are the kind of reader who wants the quotes without the context, Chapter 3 will give you plenty of ammunition. If you are the kind who wants the context without the quotes, Chapter 4 will. Reading only one of them will make you confidently wrong, and I would rather you did not use my book for that.
Two more things. First, I have tried very hard not to write "Hinduism did this." The Dharmashastras are one genre inside an enormous tradition that also produced the Buddha, Basava, Kabir and Ravidas — people who attacked this system from inside India, using Indian arguments, centuries before any European arrived. A tradition that produces both the rule and its fiercest critics cannot be summarised by either one.
Second, this is where the ground stops being speculative. From here on there are dates, documents and named people. The arguing gets harder and the evidence gets better. That is a good trade.
— Lovepreet Singh
July 2026
1From Four To Three Thousand
The four-varna scheme is the poster. Jati is the building. Here is how one became the other.
Here is a question that sounds simple and is not. If the system has four categories, why does India have thousands of castes?
Because the four were never the system. They were the theory of the system — a tidy framework that a vast, messy, multilingual subcontinent was pushed into afterwards, and never quite fitted.
What people actually lived in was jati. Thousands of them. Tied to a region, a language, a trade, sometimes a single valley. And the crucial point: your jati, not your varna, decided who you could marry, who you could eat with, and what you did for a living.
Six ways a jati gets made
Historians and sociologists have identified several mechanisms. They all operated at once, over a very long time.
- Guilds that closed. Ancient India had craft and trade associations — shrenis — of weavers, metalworkers, oil-pressers, merchants. Guilds everywhere in the pre-modern world tended to become hereditary. In India they also became marriage circles, which is the step that turned a guild into a caste.
- Tribes absorbed from outside. As settled agriculture expanded into forest and hill country, whole communities were brought into the hierarchy — almost always near the bottom. This is one of the main routes by which the system grew, and it never really stopped.
- Sects that hardened. Religious movements that began as open to all frequently became closed communities within a few generations, marrying only among themselves.
- Distance. One caste that migrated and settled far away would, after a few generations of not marrying back, become a separate jati with the same name.
- Splitting over disputes. Communities divided over ritual arguments, over whether widows could remarry, over which deity to follow — and each fragment became its own marriage circle.
- Climbing. Groups that gained land or power adopted the practices of those above them — vegetarianism, ritual, a new origin story — and pressed a claim to a higher rank. The sociologist M. N. Srinivas called this Sanskritisation.
Sanskritisation is real and well documented. Communities have moved up the ranking over generations, and many have.
But look at what kind of mobility that is. A whole community, over several generations, by adopting the customs of its superiors and successfully arguing a new origin story. It is not an individual walking into a different life. And crucially, it does not challenge the ladder — it climbs it. Every successful climb confirmed that the ranking was worth having.
Side A: caste was never the frozen block it is described as. Groups rose and fell constantly. A system with that much internal movement is not a rigid hierarchy; it is a fluid one with a ranking.
Side B: that a community had to spend three generations imitating its superiors to be treated slightly better is a description of how total the hierarchy was, not evidence against it. Mobility that requires you to accept the ranking is the hierarchy working exactly as designed.
Why this multiplication matters for 2026
Because it explains something that confuses everybody in the reservation argument.
When people say "the SC quota has been captured by a few communities," or when the Supreme Court permitted states to sub-classify Scheduled Castes in 2024, they are running into this fact: the official categories are varna-shaped, and real life is jati-shaped.
Four boxes were laid over thousands of communities with wildly different histories. Some inside a single box were far more deprived than others. That mismatch was built in at the beginning, and it is now one of the central problems of Indian public policy. Part 10 returns to it.
2The Rulebooks
What the Dharmashastras actually are — and four things about them that most people on both sides get wrong.
Somewhere between roughly 600 BCE and 500 CE, a genre of Sanskrit literature developed that set out rules for how life should be lived — duties, rituals, inheritance, marriage, crime, punishment, and the obligations of each varna.
These are the Dharmashastras. The most famous is the Manava Dharmashastra, universally known as the Manusmriti — the Laws of Manu.
Four corrections before we read a word of it
Hinduism has no single authoritative scripture, no council that fixed a canon, and no central authority that could have declared one. The Manusmriti is one text in one genre. Others in the same genre — the Yajnavalkya Smriti, the Narada Smriti, and many more — disagree with it, sometimes sharply, on inheritance, on women's property, and on punishment.
Anyone who says "Hinduism says X because Manu says X" has mistaken one book for a library.
Patrick Olivelle's critical edition and translation (2005) — the standard scholarly text — places it in that window, with numismatic evidence pointing to the second or third century CE.
That is well over a thousand years after the Rig Veda. Whatever the Manusmriti describes, it is not the original Vedic society. It is a much later attempt to codify.
Olivelle's critical edition identifies passages that appear to be interpolations — material added after the original composition. This is normal for ancient texts transmitted by hand over centuries.
It is also a real complication for both sides. Defenders sometimes use it to dismiss any verse they dislike as "an interpolation," which is not how textual criticism works. Critics sometimes ignore it entirely. The honest position is that the harshest verses are stable across the major manuscripts and translations — they are not a modern forgery — but the text did grow over time.
When the East India Company began administering Indian law, it needed a written code for "Hindu law." It commissioned translations, and Sir William Jones produced an English Manusmriti in 1794.
Colonial courts then began treating it as an authoritative legal source in a way no pre-colonial Indian state consistently had — because pre-colonial justice was mostly administered by local caste councils and regional custom, not by a single Sanskrit book.
This is one of the strongest and best-documented points the "it was not really the law" side has. It is also, as we will see in Chapter 4, less of a rescue than it first appears.
Because of what happened to it afterwards. The Manusmriti was copied, commented on, taught and quoted for close to two thousand years. It was never repudiated by any authority within the tradition, because there was no authority that could repudiate anything.
And when Indians in the twentieth century looked for a symbol of the system they were fighting, both sides picked the same book. Ambedkar publicly burned a copy at Mahad on 25 December 1927. He did not choose it at random. He chose it because everyone already knew what it stood for.
A text that both its defenders and its most famous enemy agree is the emblem of the system is not a marginal document, whatever its legal status once was.
3What Manu Actually Says
No paraphrase, no softening. The verses, as they are. Then, in the same chapter, the verses that complicate them.
Most people arguing about the Manusmriti have never read a line of it. So here it is. Translations follow the standard English versions in general circulation; the wording varies slightly between translators but the content does not.
The structure
"But for the sake of the prosperity of the worlds, he caused the Brahmana, the Kshatriya, the Vaishya and the Shudra to proceed from his mouth, his arms, his thighs and his feet."
The Purusha Sukta image again, now in a law book. And immediately after it, something the hymn never did — assigned duties.
"One occupation only the lord prescribed to the Shudra — to serve meekly these other three castes."
That is the shift. In the Rig Veda, four groups exist. Here, one of them has a single permitted function, and it is service.
The punishments
These are the verses people usually only hear about. Here they are.
8.270 — "A once-born man who insults a twice-born man with gross invective shall have his tongue cut out; for he is of low origin."
8.271 — "If he mentions the names and castes of the twice-born with contumely, an iron nail ten fingers long shall be thrust red-hot into his mouth."
8.272 — "If he arrogantly teaches Brahmanas their duty, the king shall cause hot oil to be poured into his mouth and into his ears."
There is no way to read that as a metaphor. It specifies an offence, an instrument, a measurement and a body part.
8.281 — "A low-caste man who tries to place himself on the same seat with a man of high caste shall be branded on the hip and banished, or the king shall cause his buttock to be gashed."
8.417 — "A Brahmana may confidently seize the goods of his Shudra servant; for, as that servant can have no property, his master may take his possessions."
And the verses about those outside the four
This is where untouchability appears in the text — not as a vague attitude, but as a residence rule, a dress code and a job description.
10.51 — "The dwellings of Chandalas and Shvapachas shall be outside the village… and their wealth shall be dogs and donkeys."
10.52 — "Their dress shall be the garments of the dead; they shall eat their food from broken dishes; black iron shall be their ornaments; and they must always wander from place to place."
10.54 — "By day they may go about for the purpose of their work, distinguished by marks at the king's command, and they shall carry out the corpses of persons who have no relatives; that is a settled rule."
Outside the village. Dead people's clothes. Broken dishes. Identifying marks. Corpse disposal.
Anyone who has visited a village where the Dalit basti still sits at the edge, past the main settlement, will recognise the geography.
And on women
9.3 — "Her father protects her in childhood, her husband protects her in youth, and her sons protect her in old age; a woman is never fit for independence."
2.31 — "Let the first part of a Brahmana's name denote something auspicious, a Kshatriya's be connected with power, a Vaishya's with wealth — but a Shudra's express something contemptible."
Worth noticing: caste and gender are written into the same book, by the same logic. That connection returns in Part 13.
Now the verses the other side quotes — from the same text
This is the part that almost never appears alongside the ones above, and it should.
10.64 — "If the offspring of a Brahmana by a Shudra female bears children to one of the highest caste, the inferior tribe attains the highest caste within the seventh generation."
10.65 — "Thus a Shudra attains the rank of a Brahmana, and in a similar manner a Brahmana sinks to the level of a Shudra."
This is significant and it is rarely acknowledged by critics. Manu explicitly permits varna to change across generations — in both directions. The text that is treated as the charter of permanent hereditary rank contains a mechanism for altering rank.
"In the case of theft the guilt of a Shudra is eightfold; that of a Vaishya sixteenfold; that of a Kshatriya thirty-twofold; that of a Brahmana sixty-fourfold, or a hundredfold, or even twice sixty-fourfold."
Punishment here scales upward with caste. The higher your rank, the greater your guilt for the same act — because, the text reasons, you knew better.
Book 2 also contains verses instructing a student to accept knowledge and good conduct from anyone, explicitly including people of low birth.
Defenders are right that the text is not a single-minded document of hatred. It contains provisions for mobility, it scales culpability upward with rank, and it tells students to learn from anyone. A critic who quotes only 8.270 is not describing the book.
Critics are right that these humane passages sit beside instructions to pour hot oil into a man's ears for speaking above his station, and that no amount of context makes 10.51 anything other than a residential segregation order.
And there is an asymmetry that decides it. The eightfold-guilt verse is undercut elsewhere in the same text, which shields Brahmanas from execution and from several corporal punishments. The seven-generation mobility rule requires a specific and unlikely chain of marriages. The mutilation verses require nothing but a Shudra who spoke.
My reading, labelled as mine: the humane verses are real and should be quoted. They are also, structurally, the exceptions. The spine of the text is a ranked order with graded punishments, and the exceptions bend around it rather than the other way round.
4Was Manu Ever Enforced?
The best defence of the tradition, put as strongly as it can be put. It is better than its opponents admit — and it does not go as far as its supporters need.
After the last chapter, the obvious question is whether any of it was ever actually done to anybody. This is the single most important dispute in Part 3, and it deserves a full hearing on both sides.
1. There was no mechanism to enforce it. Pre-modern India had no national legal system, no police force, no appellate structure and no Sanskrit-literate magistracy in every district. Most disputes were settled by village councils and caste panchayats applying local custom.
2. Custom formally outranked text. Indian legal tradition repeatedly acknowledges the authority of regional and community custom — desachara and jatidharma. A rule in a Sanskrit treatise did not override what a community actually did.
3. Almost nobody could read it. The Manusmriti is in Sanskrit, in a society where literacy was rare and Sanskrit literacy rarer. Most Indians who lived under caste had never heard of the book.
4. The shastras contradict each other. Other Dharmashastras differ on inheritance, women's property and punishment. A body of rules that disagrees internally cannot function as a code.
5. The evidence for the punishments is textual, not archaeological. There is no body of records showing Indian states routinely pouring hot oil into people's ears. The prescription exists; a documented history of its execution largely does not.
6. The British inflated it. Colonial administration needed one written code and manufactured a uniformity that had never existed, giving Manu an authority in the nineteenth century that it had not held in the twelfth.
How much of that survives scrutiny
Quite a lot, and I want to say that plainly before the reply, because this argument is usually dismissed rather than answered.
Points 1 through 4 are largely correct as history. Point 6 is well documented. Point 5 is a fair observation about the evidentiary record.
Now the reply.
The defence answers a charge nobody serious is making. The claim is not that Indian kings ran a nationwide programme of judicial mutilation from a Sanskrit manual. The claim is that Indian society was organised by hereditary rank, that the rank determined where you lived, whom you married and what work you did, and that the texts articulated and legitimised this. That claim does not require a single verse of Manu to have been enforced by any court.
Custom was not an escape from caste — it was the delivery system. Saying "local custom governed, not Manu" is only comforting if local custom was kinder. The caste panchayats that actually enforced things were themselves caste bodies. In many places custom was harsher and faster than any text, because it needed no trial.
Illiteracy cuts the wrong way. A person who cannot read the rule excluding them is not thereby unexcluded. The barred well does not require you to have read about it.
The physical record is not textual. Segregated settlements, separate wells, separate cremation grounds and temple exclusion are not claims about a book. They are visible facts about the built environment of Indian villages, many of which persisted into living memory and some of which persist now.
And the colonial argument has a date problem. The genetic evidence in Part 2 places the endogamy lock at roughly two thousand years ago — many centuries before any European reached India. The British hardened, catalogued and formalised. They did not start it.
The defence wins the narrow question and loses the broad one.
Narrow question — was the Manusmriti the enforced legal code of India? No. The defence is right, and people who imply otherwise are overstating.
Broad question — was Indian society organised by hereditary rank with real material consequences? Yes. And that question is not answered by anything in the defence, because it never depended on the book in the first place.
Watch for the switch. The defence is usually deployed against the narrow claim and then treated as having settled the broad one. It has not. Whether the injury was real is a matter of documentary evidence from periods we can actually see — and that evidence is the whole of Part 6.
5The Lock
Forget the texts for a moment. One single mechanism explains why caste outlived every empire that ruled over it.
If you removed every copy of the Manusmriti from the world, caste would have survived unaffected. If you removed one thing — the rule that you marry inside your group — it would have collapsed within a few generations.
Endogamy is the load-bearing wall. Everything else is decoration on it.
Why marriage is the whole system
- It makes the group hereditary. A category you can marry out of stops being a birth group in two generations. Close marriage and birth becomes destiny automatically — no enforcement required.
- It keeps property inside. Land, tools, workshops, trade contacts and credit all pass through marriage. Closed marriage means closed economies.
- It preserves the occupation. If the potter's daughter can only marry a potter, the trade stays in the community whether or not any law says so.
- It creates the purity logic. Once groups are sealed, contact between them becomes something to be regulated — food, water, touch. The rules about pollution are downstream of the rules about marriage, not the other way round.
- And it needs no state at all. This is the crucial one.
As covered in Part 2: genome-wide analysis of 73 groups across the subcontinent dated the mixture between ancestral populations to roughly 1,900–4,200 years ago. Before about 4,000 years ago there was essentially no mixing barrier. After that, mixture was widespread — and then it ceased, and endogamy became the norm.
Moorjani et al., American Journal of Human Genetics, 2013
Anuloma and pratiloma: the asymmetry
The texts do discuss marriage across varna, and the way they discuss it is revealing.
| Term | What it is | How the texts treat it |
|---|---|---|
| Anuloma "with the grain" | Higher-varna man marries lower-varna woman | Disapproved of in places, but recognised and regulated, with rules for the status of the children |
| Pratiloma "against the grain" | Lower-varna man marries higher-varna woman | Condemned in the strongest terms. Offspring assigned the lowest status. |
If the concern were purely about keeping occupations tidy, direction would not matter. It matters enormously. Which tells you the system was not really regulating work — it was regulating women, and through them, the boundary of the group.
This is why caste and gender cannot be separated in Indian history, and why the strictest controls on women's movement, remarriage and choice have historically tracked the castes most anxious about rank.
The measurement that predicts the future
Large national surveys have consistently found that the overwhelming majority of Indian marriages remain within caste, and that the figure has moved very little across decades — despite urbanisation, education and mass migration to cities.
This is the single most useful number in the entire debate, and almost nobody cites it. Reservation policy argues about entry into colleges and jobs. The lock is in the wedding.
Which produces an uncomfortable question for both sides, and I will pose it here and answer it in Part 14.
If the mechanism that sustains caste is marriage, and if reservation does not touch marriage — then what, exactly, do we expect reservation to fix? And if a movement wants to end caste, why is its programme about exam seats rather than about the one practice that keeps caste alive?
6The Five Walls
Strip away the theology. Caste was five practical rules about ordinary life — and each one did a specific job.
Ask most people what caste is and they will say something about hierarchy or discrimination. That is the effect. This chapter is the mechanism — the five actual, physical separations that made a jati a jati.
None of them requires a text. All of them are enforced by neighbours.
Wall 1 — Marriage (connubium)
Covered fully in Chapter 5. You marry inside the group. This is the wall that makes all the others possible, because it is the one that converts a category into a bloodline.
What it does: makes the group hereditary, keeps property inside, keeps the trade inside, and needs no enforcer beyond a family.
Wall 2 — Food and water (commensality)
Who may cook for you. Who may serve you. Whose water you may drink. Whether you may eat in the same row, or must eat afterwards, or outside.
This is the wall that made caste visible every single day. Marriage happens once in a lifetime. Meals happen three times a day, and every one of them was a public performance of rank.
What it does: converts an abstract ranking into a constant, unmissable social fact. And it makes the boundary self-policing — nobody has to be punished, because everyone is watching.
Of all the sites of caste conflict in modern Indian history, the most common is the water source. Not the temple, not the school — the well and the tank.
The reason is structural. Water is unavoidable, public and daily. A rule that you may not draw water is a rule you collide with every morning, in front of everyone. That is why the anti-caste movement's most famous early action, the Mahad Satyagraha of 1927, was about drinking water from a public tank — and why the response to it was so violent.
Wall 3 — Residence
Where your house stands. The Manusmriti prescribes it explicitly for those outside the four varnas: outside the village. In practice, across enormous parts of India, the settlement was and in places still is physically divided — the main village, and then the separate hamlet.
What it does: makes the hierarchy architectural. A rule written in a book can be forgotten. A rule built into the map of the village is re-taught to every child born there, without anyone saying a word.
Wall 4 — Occupation
What work you do, and — just as importantly — what work you are not permitted to do. Hereditary trade, inherited alongside the tools and the customers.
The wall works in both directions, and the upward direction is the one people forget. It was not only that certain groups were forced into degrading work. It was that they were barred from everything else. A blocked exit is more confining than a bad job.
What it does: locks economic position across generations, and ties status to a specific kind of labour so tightly that the two become the same word. Many Indian caste names are simply the name of a trade.
Wall 5 — Ritual access
Who may enter the temple. Who may hear or recite the sacred texts. Who may perform the ceremonies. Who requires an intermediary to approach the divine at all.
What it does: this is the wall that made the other four feel legitimate rather than merely enforced. If the arrangement is cosmic, then resisting it is not just illegal, it is impious. That is a far stronger lock than any punishment.
Look at what is on that list: marriage, meals, housing, work and worship. Every single one is administered by families and neighbours in the course of ordinary life.
No king is required. No army, no court, no scripture, no police. Which is exactly why every empire that ruled India — Mauryan, Gupta, Sultanate, Mughal, British — passed over the system without dislodging it. There was no lever to pull.
Count how many of the five walls modern policy actually touches.
Reservation operates on Wall 4 — occupation — by opening entry to government jobs and public education. That is real, and it is the wall most amenable to state action.
Anti-untouchability law formally addresses parts of Walls 2, 3 and 5.
Nothing at all touches Wall 1. And Wall 1 is the one holding up the other four.
So when the abolition side says reservation has not ended caste, they are correct — and they have not noticed that this is an argument about what reservation was never designed to do. And when the retention side treats reservation as the answer to caste rather than as a partial remedy for one of its effects, they are making the same category error from the opposite direction.
7Who Actually Benefited?
The answer is not the one either side gives — and getting it wrong is why the modern argument keeps missing its target.
The standard picture is simple: Brahmins at the top, everyone else below, Brahmins the beneficiaries. It is the picture used by both the anti-caste movement and by its opponents, who find it convenient for opposite reasons.
It is also, as a description of how power worked in most of India for most of history, substantially wrong.
Ritual rank and actual power came apart
Brahmins held the highest ritual rank. That is not the same as holding the land, the grain, the loans or the muscle.
Across most of rural India, the people who actually controlled village life were what sociologists call the dominant castes — communities that were not usually Brahmin, often ranked in the middle of the varna scheme, but which owned most of the cultivable land, commanded numbers, and settled disputes.
M. N. Srinivas's concept of the dominant caste describes a community that is numerically strong in a locality, owns a large share of the land, and therefore wields decisive local power — regardless of where it sits in the ritual hierarchy.
Many poor Brahmin families — village priests, cooks, scribes — held high ritual status and very little else. Meanwhile landowning communities ranked below them in theory controlled everything that mattered in practice: employment, credit, water, and the physical enforcement of the rules.
Why this matters more than almost anything else in this book
Because it explains the shape of modern Indian politics, and because it is the fact that makes both popular narratives collapse.
"Brahminical oppression" as a complete explanation does not survive contact with village history. In much of India, the people enforcing untouchability at the well, refusing to lease land, and organising violence after an inter-caste marriage were not Brahmins. They were the locally dominant landowning castes — many of whom are today classified as OBC and hold reservation themselves.
An analysis that aims exclusively upward at the ritual apex will keep missing the people who actually did the enforcing.
The picture of a small elite hoarding everything while the general category suffers also collapses. The system's beneficiaries were spread widely through the middle of the hierarchy, not concentrated at the top.
Which means the number of Indian families that gained something from caste — access to land, to credit, to labour they did not have to pay properly — is far larger than the Brahmin population. "My family never oppressed anyone" is a much bigger claim than most people making it realise.
Once you see the dominant castes, Indian caste politics stops being a two-sided fight and becomes a three-sided one.
At the ritual top: Brahmins and allied communities — high status, disproportionate presence in education and administration, small in number.
In the middle: large landowning and cultivating communities — enormous in number, historically dominant locally, ritually ranked below, and now the core of OBC politics.
At the bottom: Dalit and Adivasi communities — excluded by both of the above, and in daily life more often in direct conflict with the middle than with the top.
This is why the Mandal Commission's 27% mattered so much politically, why so much rural caste violence is between OBC and Dalit communities rather than between Brahmin and Dalit, and why demands from Marathas, Jats and Patidars — dominant communities asking to be included in quotas — are not the paradox they appear to be.
Part 9 and Part 10 are unreadable without this chapter.
Neither the version with Brahmins as sole villain nor the version with no villains at all survives the evidence. Caste distributed both privilege and harm unevenly across dozens of rungs, and a great many communities were simultaneously oppressed from above and oppressive to those below.
That is the least satisfying possible answer and the closest to the record.
8Where Untouchability Came From
Five serious theories. Ambedkar's own, and its four rivals — each at full strength, then scored.
Untouchability is not simply the bottom of the caste ladder. It is a different thing: a category of people placed outside the four varnas altogether, whose physical presence was treated as polluting.
Where did it come from? There is no consensus. Here are the five explanations that have serious support.
Set out in The Untouchables: Who Were They and Why They Became Untouchables (1948). Ambedkar argued that as settled agricultural villages formed, scattered remnants of defeated or broken tribes — "Broken Men" — attached themselves to villages as watchmen and menial workers, living outside the settlement.
Then two things happened. Buddhism spread, and Brahmanism responded by out-competing it on non-violence — abandoning animal sacrifice and adopting cow protection and vegetarianism. Beef-eating, previously ordinary, was redefined as an abomination.
The Broken Men, poor and dependent on dead cattle, went on eating beef. They became the visible embodiment of impurity. Ambedkar dated the emergence of untouchability proper to around 400 CE.
Strengths: explains the geography — why these communities live outside the village boundary, which is otherwise very hard to account for. Explains the specific link between untouchability and dead-animal work. Fits the timing of the Buddhist–Brahmanical contest.
Weaknesses: the Broken Men mechanism is an inference rather than a documented process. Ambedkar was also, openly, a partisan — which does not make him wrong, but does mean the theory needs the same scrutiny as any other.
Associated above all with Louis Dumont's Homo Hierarchicus. On this view caste is organised by a single principle — the opposition of pure and impure. Occupations dealing with death, excrement, leather and butchery are polluting; those who perform them are permanently polluted; and untouchability is simply the extreme end of a continuous scale.
Strengths: explains why the specific occupations are what they are, across the entire subcontinent, in unrelated language regions.
Weaknesses: heavily criticised for treating an ideology as though it were the whole reality, and for underweighting power, land and force. Critics argue Dumont explained how Brahmins described the system rather than how it worked.
Associated with G. S. Ghurye and others. As agrarian society expanded, it absorbed communities that had been outside it — and slotted them in at the lowest available rank, or below it.
Strengths: a documented, continuing process; explains regional variation in who is untouchable and why the lists differ so much between states.
Weaknesses: explains recruitment into the category but not the invention of the category itself.
Associated with R. S. Sharma's work on early Indian feudalism. After the Gupta period, long-distance trade contracted, cities declined, coinage became scarce, and land grants to Brahmins and temples multiplied. The economy localised. In a shrunken, land-based world, occupational groups lost mobility and hardened into hereditary service relationships.
Strengths: the timing matches the period when caste rigidity appears to intensify, and it supplies a material mechanism rather than only an ideological one.
Weaknesses: the "Indian feudalism" model has been seriously contested by other historians, who dispute both the scale of urban decline and the analogy with Europe.
Associated with colonial ethnographers, above all H. H. Risley, who attempted to correlate caste rank with nose measurements.
Strengths: historically the most influential theory, and it shaped a century of writing.
Weaknesses: as set out in Part 5 of this series and Part 2 of this book, it has been substantially abandoned. The genetic evidence does not show the clean population split it requires, and the linguistic basis for reading varna as skin colour is weak.
Theory 5 is largely out. Theory 2 explains the logic but not the origin. Theory 3 explains recruitment but not invention. Theories 1 and 4 are the strongest, and they are compatible with each other: a material collapse that froze occupational groups, plus a religious contest that turned one specific practice into a marker of permanent impurity.
Nobody has proved any of this. What can be said with confidence is narrower and still substantial: untouchability was not present in the earliest layers, it emerges in the first millennium CE, and by the medieval period it is unmistakably established.
9The First Revolt
Twenty-five centuries before any modern movement, someone stood up and said birth decides nothing. Then his religion left the country.
One of the most persistent claims in this whole argument is that opposition to caste is a foreign import — brought by missionaries, colonisers, or Western-educated reformers.
It is not, and the evidence against it is about two and a half thousand years old.
na jaccā vasalo hoti, na jaccā hoti brāhmaṇo |
kammanā vasalo hoti, kammanā hoti brāhmaṇo ||
"Not by birth is one an outcaste; not by birth is one a Brahmin. By deeds one becomes an outcaste; by deeds one becomes a Brahmin."
The Buddha rejected birth-based worth explicitly and repeatedly. Buddhist orders admitted members regardless of varna. Mahavira and the Jain tradition took a comparable position. Later came Basava and the Lingayats in twelfth-century Karnataka, and the Bhakti poets — Kabir, Ravidas, Nanak, Tukaram, Chokhamela — across many centuries and languages.
Whatever else is true, "criticising caste is a Western idea" is false, and it is false by a margin of roughly two thousand years.
The uncomfortable follow-up
Buddhism rejected caste, spread across Asia, and then very largely disappeared from the land where it began.
Why? There is no single answer, and historians treat it as a centuries-long process rather than an event. The main factors, as generally set out:
- Loss of royal patronage. Buddhism depended heavily on kings and wealthy donors. After the Gupta period, patronage shifted increasingly toward Brahmanical Hinduism and its temples.
- Absorption. Hinduism took in a great deal — non-violence, compassion, meditative practice — and eventually incorporated the Buddha himself as an avatar of Vishnu. A rival you have absorbed is no longer a rival.
- Brahmanical revival. Shankara in the eighth century and the Bhakti movements offered philosophical rigour and emotional devotion in the same package.
- Monastic distance. Buddhism concentrated in large monasteries, while Brahmanical religion was embedded in household rituals — birth, marriage, death. The tradition that owns the wedding owns the society.
- Destruction of the great centres. Turkic invasions in the twelfth and thirteenth centuries destroyed Nalanda, Vikramashila and Odantapuri. Nalanda is generally dated to around 1193. For an institution-based religion, this was close to fatal.
For those who say caste is not central to Hinduism: the tradition's most serious internal challenger for a thousand years was a movement whose founding position included the rejection of birth-based rank — and it lost. That is a fact about what the society chose.
For those who say Hinduism is simply a caste ideology: the fiercest attacks on caste in world history were made by Indians, in Indian languages, using Indian religious arguments, for two and a half millennia. A tradition that keeps generating its own abolitionists is not reducible to the thing they are abolishing.
And the mechanism from Chapter 5 explains the outcome better than either. Buddhism offered a philosophy. Caste was not held in place by philosophy. It was held in place by who your daughter could marry — and the monasteries had nothing to say about that.
10The Hardening
Fifteen hundred years, on one timeline. Watch a metaphor turn into a wall.
11Nine Questions Nobody Asks
The blind spots in this stage of the argument — including two that ought to worry both sides equally.
1. Why does everyone argue about a book almost nobody obeyed?
Both sides have an interest in the Manusmriti being central. Critics get a document with quotable horrors. Defenders get a target they can plausibly discredit as unenforced and colonially inflated.
What both avoid is the harder subject: the actual, undocumented, everyday operation of jati — enforced by neighbours, not by kings. There is no book to burn and no book to defend. Which is precisely why the argument stays where the book is.
2. If custom governed rather than text, who was making the custom?
"It was local custom, not Manu" is offered as a defence. But local custom was made and enforced by caste panchayats, and those bodies were the caste system organising itself.
Saying the rules came from the community rather than from a book does not make the rules milder. It makes them harder to appeal.
3. Why did untouchability attach to work with dead things — in every language region at once?
Across an enormous, linguistically fragmented subcontinent, the same occupations turn up at the bottom: handling corpses, leather, butchery, human waste. Communities with no contact and no shared language arrived at nearly the same ranking.
Something extremely powerful and extremely portable was doing that. Whatever theory you prefer has to explain the uniformity, not just the existence.
4. Why is there no equivalent of an abolition?
Slavery was abolished — by acts, dates, wars, proclamations. Serfdom was abolished. Apartheid ended on a date.
Caste has no such moment. The Constitution abolished untouchability in 1950, and untouchability is a subset. There was never a point at which Indian society collectively renounced the ranking itself. The absence of a founding renunciation matters as much as the absence of a founding decree.
5. Does "it was never enforced" actually help the defence?
Consider what that argument concedes. If a hierarchy needed no laws, no police and no punishments to sustain itself for two thousand years, it was more deeply held than one requiring enforcement — not less. A system maintained voluntarily by millions of families is a harder thing to condemn a state for, and a harder thing to end.
6. Whose fault is it if nobody enforced it?
This follows directly and it is genuinely uncomfortable. A hierarchy imposed by conquerors has a villain and a plan for removal. A hierarchy maintained by ordinary people, in their own homes, about their own children's marriages, has neither — and no innocent bystanders. Which is why this framing is unpopular with everybody.
7. Why do reformers keep getting absorbed?
The pattern repeats with remarkable consistency. A movement rejects caste. It attracts followers. Within a few generations its followers have become a caste — endogamous, ranked, with a place in the local order.
It happened to Buddhist communities, to Lingayats, to several Bhakti sects, and to some Sikh and Christian and Muslim communities in India. The system does not usually defeat its challengers. It digests them.
8. What did caste do that was useful? And why does asking feel forbidden?
Uncomfortable, but serious. Jati provided credit networks, insurance, skills transmission, dispute resolution and migration support in a society with no state welfare and no banks. For many people it was the only safety net that existed.
This is not a defence — systems can be functional and unjust at once, as feudalism was. But if you cannot explain what a system did for people, you cannot explain why they kept it, and "everyone was brainwashed for two thousand years" is not an explanation.
9. If marriage is the lock, why is the entire national argument about jobs?
The mechanism that creates and sustains caste is who marries whom. Reservation operates on college admissions and government employment. These are not the same lever.
Neither side finds this convenient. Supporters must explain how a policy aimed at employment addresses a structure held together by marriage. Opponents must explain why a movement that claims to want caste gone campaigns about exam cut-offs rather than the practice that reproduces it.
12What The Evidence Can And Cannot Settle
The honest scorecard for Part 3.
| Question | Status | Best current answer |
|---|---|---|
| Does the Manusmriti prescribe graded punishment by caste? | Settled | Yes. The verses are stable across translations. |
| Does it also contain mobility and humane provisions? | Settled | Yes — 10.64–65 permits varna change over generations |
| Was it the enforced legal code of India? | No | Custom and caste councils governed, not a Sanskrit text |
| Did the British inflate its authority? | Yes | Documented, from 1794 onward |
| Was Indian society organised by hereditary rank? | Yes | And this never depended on the text |
| When did endogamy become the norm? | Datable | Roughly 1,900–4,200 years ago |
| Where did untouchability come from? | Open | Five theories; two strongest and compatible |
| Is anti-caste thought a foreign import? | No | Indigenous and continuous for ~2,500 years |
| How severe was it in practice? | Answerable | But only from documents — Part 6 |
The three things nobody can tell you
- How much of the text was ever acted on. We have the prescriptions. We do not have court records from the period. The gap between what was written and what was done is genuinely unmeasurable at this distance.
- What it felt like from below. Again: the archive belongs to the literate. We have Chokhamela and Ravidas, and they are precious precisely because they are so rare.
- Why the lock closed. Still unanswered, and still the most important open question in Indian history.
Part 2 ended by noting that the origin of caste barely affects the case for a remedy. Part 3 changes that, in one specific way.
We now have a mechanism, and mechanisms have implications that origin stories do not.
If caste is sustained primarily by endogamy — by millions of private decisions inside families, with no central enforcer — then several things follow immediately, and they are inconvenient in different directions.
Inconvenient for the abolition side: removing state recognition of caste will not touch the mechanism. Caste does not need the state. It never did. It survived empires that had no interest in maintaining it. It will comfortably survive the deletion of a column on a form.
Inconvenient for the retention side: a policy that operates on jobs and admissions is not aimed at the mechanism either. It can produce representation, income and visibility — real goods. It cannot, by itself, unlock the lock.
Both movements are working on the symptoms. The difference is that one of them is at least treating something.
13The Five-Line Nichod
Everything in this book, compressed. If you remember nothing else, remember these.
PART 3 — HOW VARNA BECAME JATI
- Varna is the poster; jati is the building. Four theoretical categories were laid over thousands of real communities formed by guilds closing, tribes being absorbed, sects hardening and groups splitting. That mismatch is why the 2024 sub-classification verdict happened.
- The Manusmriti says what critics say it says — and also more than they admit. It prescribes mutilation for insolence and residence outside the village for those beyond the four varnas. It also permits varna to change over seven generations and scales guilt upward with rank. Print all of it or none of it.
- It was not the enforced law of India — and that rescues less than it seems. Custom and caste councils governed, and the British inflated the text's authority from 1794. But the claim that Indian society was organised by hereditary rank never depended on the book.
- Endogamy is the whole machine. Marriage inside the group makes caste hereditary, keeps property in, preserves occupation, generates the purity rules — and requires no state at all. That is why caste outlived every empire that ruled over it, and why inter-caste marriage rates remain in the low single digits today.
- Anti-caste thought is Indian and it is 2,500 years old. The Buddha: "not by birth is one an outcaste." Then Basava, Kabir, Ravidas, Nanak, Tukaram. The claim that opposing caste is a Western idea is false by roughly two millennia — and the fate of those movements shows the system usually digests its critics rather than defeating them.
Coming next
A claim you have definitely heard, tested properly. Caste under the Delhi Sultanate and the Mughals: what changed and what did not. Caste inside Indian Islam and Indian Christianity — Ashraf, Ajlaf, Arzal, and Dalit Christians — a set of facts inconvenient for several sides at once. Did conversion actually let anyone escape? And the Bhakti movement in full: Kabir, Ravidas, Nanak, Tukaram, Chokhamela — India's own anti-caste tradition, in vernacular languages, reaching the people the Sanskrit texts never touched.
A closing thought: this was the hardest part to write fairly, and I expect it to be the hardest to read fairly. If you got through Chapter 3 and Chapter 4 and felt the pull of both, you are reading it correctly. — L.S.
Sources & further reading — Part 3
Every factual claim in Part 3, and where it comes from.
The Manusmriti
Patrick Olivelle, Manu's Code of Law: A Critical Edition and Translation of the Mānava-Dharmaśāstra (2005) — the standard scholarly edition; dating to c. 200 BCE – 200 CE, with numismatic evidence suggesting the second or third century CE; identification of probable interpolations. · Verse translations follow the standard English renderings in general circulation, principally Bühler's (Sacred Books of the East, vol. 25); wording varies between translators but the substance of the verses cited does not. Verses quoted: 1.31, 1.91, 2.31, 8.270–272, 8.281, 8.337–338, 8.417, 9.3, 10.51–52, 10.54, 10.64–65. · Sir William Jones's English translation, 1794, and its use by colonial courts as an authority for "Hindu law."
Endogamy and genetics
Moorjani et al., "Genetic Evidence for Recent Population Mixture in India," American Journal of Human Genetics, 2013 — mixture dated to approximately 1,900–4,200 years ago, followed by the shift to endogamy. Full discussion in Part 2 of this series. · Contemporary inter-caste marriage rates: consistently reported in the low single digits across large national survey datasets.
Untouchability
B. R. Ambedkar, The Untouchables: Who Were They and Why They Became Untouchables (1948) — the Broken Men thesis; the link between Brahmanical cow protection, the redefinition of beef-eating and the emergence of untouchability; Ambedkar's dating to around 400 CE. Recently reissued in annotated selection as Beef, Brahmins, and Broken Men (Columbia University Press). · Louis Dumont, Homo Hierarchicus — the purity/pollution model, and the substantial body of criticism of it. · G. S. Ghurye — tribal absorption. · R. S. Sharma on early Indian feudalism — post-Gupta urban decline, land grants and the localisation of the economy; note that this model is contested by other historians. · H. H. Risley — the racial theory, now substantially abandoned. · Mahad, 25 December 1927 — the public burning of the Manusmriti.
Buddhism, Jainism and the Bhakti tradition
Vasala Sutta, Sutta Nipata 1.7 — "not by birth is one an outcaste… by deeds one becomes a Brahmin." · On the decline of Buddhism in India, historians treat it as a centuries-long process with multiple causes rather than a single event: loss of royal patronage after the Gupta period; absorption of Buddhist ideas and of the Buddha himself into the Vaishnava avatar scheme; the Brahmanical revival associated with Shankara in the eighth century; the concentration of Buddhism in monasteries rather than household ritual; and the destruction of Nalanda, Vikramashila and Odantapuri in the Turkic invasions, Nalanda generally dated to c. 1193. · Basava and the Lingayat movement, twelfth-century Karnataka. · M. N. Srinivas — Sanskritisation.
A note on what is missing — and on corrections
Part 3 makes no claim about how severely caste rules were applied to real people in periods we can document, about the medieval Islamic period, about colonial administration, or about conditions today. Those are Parts 4, 5, 6 and 11. Anything marked CONTESTED CLAIM stays contested until evidence is produced — including claims made by people I might agree with.
Sanskrit and Pali transliteration follows standard conventions. Verse numbering in the Manusmriti varies slightly between editions; the numbering used here follows the common Bühler sequence. If you find an error, I want to know. Corrections will be carried in later parts.