The Part Where I Was Wrong
Writing Part 1 was uncomfortable. Writing Part 2 was worse.
I came into this section believing I already knew the answer. I had heard the standard line my whole life — that varna was originally a clean division of work, that it had nothing to do with birth, and that the whole ugly business of caste came later, probably from foreigners. It is a comfortable story. It lets you love your civilisation without flinching.
Then I went and read the actual verses. Not summaries of them. The verses.
And what I found was messier than either side had told me. Some of what I believed turned out to have real support — better support than I expected, and better than its critics usually admit. And some of it fell apart completely the moment I checked the dates.
So this part is built differently from Part 1. In Part 1 I could often say "this is simply wrong" because there was a rule or a court judgment to check it against. Here there is no judge. There are hymns written down thousands of years ago in a language almost nobody now speaks, a city whose writing we cannot read, and DNA that tells us who married whom but never tells us why.
That means this part has more yellow boxes than green ones. That is not me dodging. It is what honest reporting on this evidence looks like. Anyone who tells you the origin of caste is a settled matter is either selling something or has not read enough.
What I can promise is this: every important verse in this book is printed in full, not paraphrased. You will see the actual words and then you will see what each side does with them. If my reading annoys you, you will have the raw material to build a better one.
One warning. This chapter touches the Aryan question, which in India is not really a history debate at all — it is a political one wearing history's clothes. I have tried to report what the evidence shows and, just as importantly, what it does not show. Both of those matter.
— Lovepreet Singh
July 2026
1The Question Under The Question
Why an argument about a hymn from three thousand years ago decides an argument about medical college seats in 2026.
Somebody reading Part 1 might reasonably ask: why are we going back this far? The fight is about exam seats. What does the Rig Veda have to do with NEET?
Everything, as it turns out. And here is why.
Reservation is a remedy. Every remedy assumes an injury. So the entire policy rests on a claim about history: that certain communities were systematically held down for a very long time, and that this is why they start behind today.
Which means the argument secretly depends on a question about the past. And there are only four possible answers to it.
| If the truth is… | …then the injury was | …and reservation looks like |
|---|---|---|
| A. Caste was rigid and cruel from the very beginning | Ancient, deep, and built into the culture itself | A minimum correction, probably still insufficient |
| B. Varna started as flexible work, then hardened over centuries | Real, but a corruption rather than an original design | Justified — but so is the argument that it can be undone |
| C. Caste rigidity came mainly from outsiders — invaders, colonisers | Real, but not the fault of Indian society itself | Justified against history, but oddly aimed |
| D. The whole thing is exaggerated or invented later | Small or imaginary | A grievance industry with no basis |
Look at that table closely, because it explains the entire internet.
Almost every angry argument you have ever seen about reservation is really two people who have silently picked different rows and are now shouting about seats. They are not disagreeing about NEET. They are disagreeing about 1500 BCE and do not know it.
What this part will and will not do
This part covers the deepest layer: the very beginning. It answers one narrow question — at the earliest point we can see, was Indian society already divided by birth?
It does not cover Manusmriti, untouchability, the Islamic period, the British, or anything modern. Those are Parts 3, 4 and 5. Trying to do all of it at once is exactly how these conversations turn into noise.
Archaeology. Objects, buildings, burials, city layouts. Tells us how people lived. Tells us almost nothing about what they believed, and nothing at all about what they called each other.
Genetics. Ancient and modern DNA. Tells us with real precision who had children with whom, and when that stopped. Tells us nothing about why.
Texts. The Vedas, the Upanishads, the epics. The only source that gives us words. But they were composed by a specific, literate, priestly group — so they tell us what that group thought, which is not the same as what everybody did.
Each source is blind where the others can see. Anyone using only one of them is guaranteed to reach a confident wrong answer.
2The First Indians
Before any texts, before any cities — who was actually here. And the single most important date in this entire book.
In the last fifteen years something happened that changed this argument permanently, and most people fighting about caste online have not caught up with it. We learned to read ancient DNA.
Not opinions about ancient people. Their actual genomes, pulled out of bones and teeth thousands of years old. It does not care about politics. It cannot be argued with by quoting a different verse.
Who Indians descend from
Modern South Asians are, overwhelmingly, a mixture of three ancient streams of people.
1. The First Indians (AASI). Descendants of the earliest modern humans to reach the subcontinent, tens of thousands of years ago. They have no close living relatives anywhere outside South Asia. Every Indian alive today carries some of this ancestry.
2. Iranian-related farmers. A population related to — but split very early from — ancient Iranians. They mixed with the First Indians, and that mixture is the population of the Indus Valley Civilisation.
3. Steppe pastoralists. Herders from the Central Asian grasslands, whose ancestry appears in South Asia in the second millennium BCE — after the great Indus cities were already old.
The Rakhigarhi skeleton
In 2019 a team published the genome of a single individual buried at Rakhigarhi in Haryana — one of the largest Indus Valley sites. It is, so far, the only Indus Valley genome we have.
The result was striking, and both sides seized on it within hours.
The genome was a mixture of Iranian-related ancestry and First Indian (AASI) ancestry — and contained none of the Central Asian Steppe component found in many South Asians today. Published in Cell, 2019, by Shinde, Narasimhan and colleagues.
In the same year a much larger study in Science analysed ancient DNA from 523 individuals across Central and South Asia and found the same pattern: Steppe-related ancestry enters the subcontinent after the mature Indus period.
Now the date that matters more than any other
Here is the finding that should be printed on the first page of every argument about caste, and almost never is.
In 2013, researchers analysed genome-wide data from 73 groups across the subcontinent and worked out when different ancestral populations stopped mixing with each other. Because when two populations mix and then stop, their DNA carries a clock.
The study found mixture dates between roughly 1,900 and 4,200 years ago. Before about 4,000 years ago, there was essentially no mixture between these ancestral groups. After that, mixture was rampant — affecting almost every group in India, including isolated tribal populations.
And then it stopped. India transformed from a place where mixing between populations was the norm into a place where marrying strictly within your own group was the norm.
Moorjani et al., American Journal of Human Genetics, 2013
Read that again, because it is the closest thing to a hard scientific date for the beginning of caste as a closed system that anyone has ever produced.
Endogamy — marrying only inside your group — is the lock on the caste system. Everything else about caste is enforceable only because of it. And we can now see, in the genome, roughly when the lock clicked shut.
For the "it was always like this" reading: the lock closes around two thousand years ago at the latest. That is roughly a thousand years before any Islamic ruler and about eighteen hundred years before the British. Whatever caused caste rigidity, it was already finished long before any foreigner blamed for it arrived. Caste is indigenous and it is old.
For the "it hardened later" reading: before roughly four thousand years ago there was no mixing barrier at all, and for a long stretch after that mixture was everywhere. Which means the earliest Vedic period was not an endogamous society. Whatever varna meant when it was first written down, it did not yet mean a closed breeding group. The rigidity came later, and it came gradually.
Both of those are true. They are not in conflict. They only look like opposites because each side quotes one and hides the other.
A deep-time timeline
It has not, and this claim gets made in both directions. What genetics shows is when populations stopped mixing. It cannot tell you whether the people at the time called that arrangement varna, or jati, or anything at all. It cannot tell you whether the barrier was religious, occupational, regional, tribal or linguistic.
DNA records the fact of separation with real precision. It has absolutely nothing to say about the reason for it. Anyone who tells you the genome proves a motive is going beyond their evidence.
3The City With No King
A hundred years of excavation in the Indus Valley. What they found — and the thing they never found.
Between roughly 2600 and 1900 BCE, the largest civilisation on earth by area was here. Bigger than Egypt. Bigger than Mesopotamia. Cities of tens of thousands of people, spread across a million square kilometres.
And it is the strangest ancient civilisation we know of, for one reason.
What they found
- Planning that borders on obsessive. Streets on a grid. Covered drains running under them. Bricks made to a standard 1:2:4 ratio across sites hundreds of kilometres apart.
- A single system of weights. Accurate, standardised, used from Gujarat to Afghanistan. Someone was enforcing a standard.
- Long-distance trade. Indus seals turn up in Mesopotamia. Mesopotamian records refer to a trading land called Meluhha, widely identified with the Indus region.
- Extraordinary public water works. The Great Bath at Mohenjo-daro. Wells across the cities. Private bathrooms in ordinary houses at a time when most of the world had nothing of the kind.
And here is what they did not find
After a century of excavation across hundreds of sites, clear evidence of a ruling class has not materialised. There is nothing resembling the pyramids, nothing like the royal tombs of Ur, nothing like the temple complexes of Mesopotamia.
As one classic summary put it: elsewhere, wealth and effort went into temples for gods and tombs for kings. In the Indus cities, the finest structures were built for the convenience of the citizens.
Even the famous "Priest-King" statuette from Mohenjo-daro got that name from the archaeologists who dug it up, based on nothing but a guess about what a ruler ought to look like. A major 2020 review in the Journal of Archaeological Research was actually titled "Killing the Priest-King" — arguing that the figure and the assumptions behind it should be retired.
The honest fight among archaeologists
This is not a case of experts agreeing and politicians disagreeing. Archaeologists genuinely disagree, and both positions are respectable.
A century of searching has produced no ruling class, and at some point absence of evidence becomes meaningful. The Indus was complex without being hierarchical — proof that you do not need a king or a priestly elite to run a sophisticated society.
The public works point the same way. Drains and wells for ordinary houses are not what a society builds when a small elite is capturing the surplus.
Burials are where hierarchy usually shows up most clearly, and Indus burials found so far are very few relative to the population. We may simply be missing the elite graves.
There are also real differences in house sizes, in access to wells, and in the citadel-versus-lower-town layout at several sites. And a civilisation that enforced identical brick ratios across a million square kilometres was clearly enforcing something.
The wall we cannot climb
We have thousands of inscribed seals. Nobody can read a single one. There is no bilingual key, and most inscriptions are extremely short.
This is the hard limit on everything in this chapter. For Egypt and Mesopotamia we can read what people said about themselves — their laws, their complaints, their tax records. For the Indus Valley we have objects and silence.
Neither statement is supportable, and both get made constantly.
The first treats "we found no kings" as "we have proved there was equality." The second reads hierarchy into differences in house sizes that could reflect a dozen things.
The truthful position is uncomfortable and short: we cannot read their writing, so we do not know what they called each other. Anyone claiming otherwise is filling silence with their own preference.
4The Argument About Arrival
The most politically explosive question in Indian history. Three positions, each written by its own best advocate.
There is no way to write this chapter that will not annoy somebody, so let me say at the top exactly what I am doing. I am going to lay out three positions, give each the strongest case its supporters make, and then report what the evidence currently supports and how confident anyone should be.
First, why this matters so violently.
If a group of outsiders arrived and imposed a hierarchy on the people already here, then caste has an origin story that looks like conquest — and the anti-caste movement gains a founding narrative of a native population subjugated by newcomers.
If nobody arrived, then Indian civilisation is continuous and indigenous, and the conquest story collapses.
Both sides therefore need a particular answer. That is precisely why you should distrust confident people here — including me.
Position 1: The Aryan Invasion Theory
The oldest version, and the one taught in Indian schools for decades. Fair-skinned Aryans swept in on chariots around 1500 BCE, destroyed the Indus cities, subjugated the dark-skinned natives, and turned them into Shudras.
No archaeological evidence of a violent conquest of the Indus cities has been found — no burnt layers of the right date across sites, no mass graves of a battle. The cities appear to decline over generations, with climate change and the shifting of river systems now the leading explanations.
The timing is also wrong: the decline of the Indus cities was well under way before Steppe ancestry shows up in the genetic record.
It matters that this was overturned by ordinary archaeology. When people say "the invasion theory was a colonial construction," that particular sentence is largely correct.
Position 2: The Indo-Aryan Migration model
Not an invasion but a long, gradual movement of pastoralist groups from the Central Asian steppe into the north-west of the subcontinent over centuries, mixing with the people already there.
Genetics. Steppe-related ancestry is absent from the Rakhigarhi Indus genome and from Indus-related individuals at connected sites, and present in much of the modern South Asian population. Something arrived between those two points.
Language. Sanskrit is unmistakably related to Greek, Latin, Persian, Lithuanian and English. Those languages form one family. Either they all spread out from somewhere, or they all spread out from India — and the second requires explaining why the deepest branches of the family sit outside India.
The Mitanni. In what is now Syria, a fourteenth-century BCE treaty invokes gods named Mitra, Varuna, Indra and the Nasatyas, and a horse-training manual uses Indo-Aryan numerals. Indo-Aryan speech was present in West Asia at that date — which fits a spreading language family.
The horse. Horse remains are essentially absent from mature Indus sites and become common later. The Rig Veda is saturated with horses and chariots.
Position 3: Out of India
The Indo-European languages originated in the subcontinent and spread outward. Nobody significant came in.
Cultural continuity. The archaeological record of the north-west shows no sharp break at the expected date. Pottery, settlement patterns and burial customs shade into each other. Migrations of the scale claimed usually leave a clearer mark.
The texts remember no homeland. The Rig Veda describes the rivers of the Punjab in loving detail, including the Sarasvati, and shows no memory of having come from anywhere else. Compare that with genuinely migratory peoples, who almost always preserve an origin story.
The Sarasvati problem. The Rig Veda praises the Sarasvati as a mighty river. Palaeo-channel and geological work suggests the relevant river system was drying out early. If the hymns describe it while it still flowed, the composers were here earlier than the standard chronology allows.
Ancestry is not identity. A genetic component arriving does not prove that language, religion or social structure arrived with it. Populations mix constantly without their culture changing hands.
Where the evidence currently stands
Among geneticists, archaeologists and linguists publishing peer-reviewed work, the movement of Steppe-related ancestry into South Asia in the second millennium BCE is treated as one of the better-established findings of the ancient DNA revolution. Out of India remains outside that consensus.
But — and this is not a small "but" — the scale, timing and social consequences of that movement remain genuinely open. Consensus on "some people moved" is not consensus on "and they imposed caste."
Suppose migration is correct. It still does not follow that the newcomers created caste. Steppe ancestry is spread across nearly all Indian populations, including many communities at the bottom of the caste order. If a migrating group had installed itself as a hereditary elite from day one, you would expect a much cleaner genetic split than we actually see. The pattern looks more like mixing that later froze than like conquest that was never undone.
Suppose Out of India is correct. That is worse for the argument it is usually deployed to support. If nobody came, then nobody foreign can be blamed for varna. It would be entirely home-grown.
This is the part almost nobody notices: the Aryan question is fought as though it decides the caste question, and it very nearly does not decide it at all.
5What "Varna" Actually Meant
One word, several meanings — and a nineteenth-century mistranslation that shaped a century of Indian politics.
Varna is a Sanskrit word. It literally means colour. It also means covering, appearance, outward form, kind, sort, class, and quality.
That one word carrying that many meanings has caused more confusion than almost anything else in this argument.
The reading that dominated for a century
When European scholars first worked through the Rig Veda in the nineteenth century, they found references to arya varna and dasa varna. They translated varna as colour, in the racial sense, and concluded that the hymns described light-skinned Aryans dominating dark-skinned natives — with the four varnas as a racial hierarchy frozen into religion.
This reading was enormously influential. It shaped colonial ethnography, it entered Indian textbooks, and it was adopted by parts of the anti-caste movement as a powerful account of origins: caste as race, imposed by conquest.
The main reasons are straightforward. Varna carries many senses in Sanskrit and the colour sense is often metaphorical — light and dark, order and chaos — as it is in many ancient languages. The Rig Veda's descriptions of dasas mix physical, linguistic and ritual differences rather than presenting a clean racial category. And the genetic record does not show the neat colour-coded population split the theory predicts.
I want to sit on this for a second, because of who it inconveniences.
This correction cuts against a framing that has been widely used by the anti-caste side — caste as fossilised racism. If you are broadly sympathetic to reservation, this is a place where the popular version of your history has been revised, and honesty requires saying so.
What the correction establishes: varna almost certainly did not begin as skin-colour racism, and the colonial race theory was bad scholarship.
What the correction does not establish: that varna was harmless. A hierarchy does not need to be racial to be a hierarchy. Removing race from the story removes one explanation for ranking. It does not remove the ranking.
This distinction gets deliberately blurred. You will often see "the racial theory of caste has been debunked" presented as though it means "the hierarchy has been debunked." It does not, and the people making the slide usually know it.
The other word: jati
You cannot understand any of this without separating two words that get used interchangeably in English and mean quite different things.
| VARNA | JATI | |
|---|---|---|
| Meaning | Colour, class, order, kind | Birth, kind, community |
| How many | Four (plus those outside the scheme) | Thousands |
| Where it lives | Mainly in texts and theory | In actual daily life — marriage, food, work |
| What it does | Provides a theoretical ranking | Determines who you marry and what you do |
Varna is the theory. Jati is the reality. When someone says "the caste system," they usually mean jati; when they defend "varna vyavastha," they are defending the theory. That gap is where a lot of arguing happens without anybody noticing they have changed the subject.
How the four became the thousands is Part 3.
6The Hymn
Rig Veda 10.90. The most consequential twelve lines in Indian history, printed in full and read by both sides.
Everything in this argument eventually arrives at one hymn — the Purusha Sukta. It describes a cosmic being, Purusha, sacrificed by the gods, and the universe made out of the pieces. Verses eleven and twelve ask what became of his body.
Here they are.
yat puruṣaṃ vyadadhuḥ katidhā vyakalpayan |
mukhaṃ kim asya kau bāhū kā ūrū pādā ucyete ||
"When they divided the Purusha, into how many parts did they arrange him? What was his mouth? What his two arms? What are his thighs and his feet called?"
brāhmaṇo'sya mukham āsīd bāhū rājanyaḥ kṛtaḥ |
ūrū tad asya yad vaiśyaḥ padbhyāṃ śūdro ajāyata ||
"The Brahmin was his mouth; the Rajanya was made from his arms; his thighs became the Vaishya; from his feet the Shudra was born."
That is it. That is the foundation stone. Four groups, arranged from the head downward, described as coming out of a single body.
Three facts about this hymn that change how you read it
The Rig Veda contains over a thousand hymns and roughly ten thousand verses. In all of that, the word śūdra occurs in this one place. This is not a system the text talks about constantly. It is a scheme that surfaces once, at the end.
The Rig Veda was not composed all at once. Scholars have long distinguished older "family books" from later additions on grounds of language, metre and content, and Mandala 10 is generally placed at the late end.
So the four-varna scheme is not the opening statement of the oldest Indian text. It appears in its most recent section.
Read the verses again. There is no instruction that a Brahmin's son must be a Brahmin. Nothing about who may marry whom. Nothing about pollution or touch. Nothing about punishment.
All of that machinery is real, and it comes later. It is the subject of Part 3. It is not in this hymn.
Now the two readings, in full
This is an image of society as an organism. A body needs a head to think, arms to protect, thighs to carry, feet to stand on. None of them is optional. The metaphor is about interdependence, not rank — a body cannot discard its feet and survive.
Nothing here says a Brahmin's child is a Brahmin. Nothing here forbids anyone from moving. It is a picture of four kinds of necessary work, in a text where the word Shudra appears exactly once — hardly the language of an obsession.
And it sits in the newest layer of the collection, which suggests a scheme being introduced rather than one that had always governed life. Later generations turned a metaphor into a prison. That is a crime committed by the interpreters, not by the hymn.
Look at the anatomy. Mouth, arms, thighs, feet. This is not a circle of equals; it is a vertical line, and everybody who has ever read it has understood which end is which. There is a reason no one describes the Brahmin as the feet.
And the source of the ranking is what matters most. This is not "some people do this work." It is a claim that the division was produced in a cosmic sacrifice — built into the structure of reality by the gods themselves.
That is the most dangerous possible framing for inequality, because an arrangement that comes from the gods cannot be reformed by people. Whether or not the hymn mentions heredity, it hands anyone who wants a permanent hierarchy the perfect instrument. And for two thousand years, that is exactly what it was used as.
Reading A is right that the text does not state heredity, and right that it appears late and rarely. People who say the Rig Veda establishes the caste system are overstating what is on the page.
Reading B is right that the ordering is unmistakably vertical and divinely sourced. People who say it is "just about division of labour" have to explain why the labour is arranged from mouth to feet and not, say, in a list.
What I think the evidence actually supports is narrower than either: the hymn is a ranking, stated as divine, without a stated rule of inheritance. It is the blueprint, not yet the building. The building goes up in Part 3 — and that is where the argument should really be happening.
7The Counter-Verses
The passages that complicate the simple story — a bard, a servant's son, a king who became a sage, and a slave woman's child.
If varna had always been strictly hereditary, certain passages in the earliest texts should not exist. They do exist. Here are the four that matter most, and what each side makes of them.
1. The family of mixed trades
"I am a bard, my father is a physician, my mother grinds corn with stones. With different thoughts, all seeking wealth, we run after our desires like cattle."
One household, three occupations, across two generations, described without a hint of scandal.
What Side A says: this is decisive. Occupation was not inherited in the Rig Vedic period. A poet whose father is a doctor and whose mother does manual grinding is a family that a rigid hereditary system could not contain.
What Side B says: none of the three occupations named is one of the four varnas, and the hymn tells us nothing about whether this family could marry outside its group or perform priestly ritual. Flexibility in trade is not the same as flexibility in status.
2. Satyakama Jabala, the boy who did not know his father
In the Chandogya Upanishad (4.4), a boy called Satyakama wants to study with a teacher and asks his mother for their lineage. She tells him honestly that she does not know — she was a servant who moved among many households in her youth, and she cannot name his father.
He goes to the teacher, Gautama Haridrumata, and repeats exactly that. The teacher's reply is one of the most quoted lines in Indian literature: only a Brahmin would speak so truthfully. He accepts him as a student.
Side A: the criterion for Brahminhood, in the text itself, is truthfulness — not birth. Explicitly not birth, in a case where birth was unknown.
Side B: the story is remarkable precisely because it is remarkable. Texts record exceptions, not routines. And the teacher's line concedes the norm even while breaking it: the boy is admitted by being declared a Brahmin, not by the category being declared irrelevant.
3. Vishvamitra, the king who became a sage
Vishvamitra is born a Kshatriya king and, through immense effort, becomes a Brahmarishi — and is credited in tradition with composing much of Mandala 3 of the Rig Veda, including the Gayatri Mantra.
Side A: the single most sacred verse in Hinduism is attributed to a man who was not born a Brahmin. Movement between varnas is not a modern reinterpretation; it is inside the tradition's own account of itself.
Side B: the length and violence of that story is the point. Vishvamitra's transformation takes lifetimes of austerity and a long feud with Vasishtha. If varna were open, none of that would be necessary. The story is evidence that the wall was real — it just shows one man scaling it.
4. Kavasha Ailusha, the son of a dasi
In the Aitareya Brahmana, a seer named Kavasha Ailusha — described as the son of a dasi, a servant or slave woman — is driven out of a sacrifice by other rishis on account of his birth. He composes a hymn, waters appear, and he is accepted back.
Side A: a man of the lowest possible birth is a recognised composer of Vedic hymns. The category was porous.
Side B: read the first half of the story. He was thrown out because of his birth. That is birth-based exclusion, recorded in the tradition's own scripture, with no apology attached. The happy ending does not delete the expulsion.
Side A's trap: collecting exceptions and presenting them as the rule. Four remarkable stories across a vast literature do not describe an open society; they describe a society where such cases were worth recording.
Side B's trap: treating later rigidity as though it were always present. These passages exist. In a fully closed system from the beginning, they could not — the texts would not preserve a slave woman's son as a hymn-composer at all.
Put the two together and you get the reading the evidence actually supports: an early system with real boundaries and real exceptions, which later lost the exceptions. That is not a compromise for the sake of politeness. It is what happens when you read all four stories instead of the two that suit you.
8The Gita's Sentence
Nine Sanskrit words, quoted in every argument about caste for a hundred years — usually only half of them.
If you have ever argued about caste with anyone, you have met this line. It is the single most cited verse in defence of the position that varna was about aptitude rather than birth.
chāturvarṇyaṃ mayā sṛṣṭaṃ guṇa-karma-vibhāgaśaḥ
"The fourfold order was created by Me according to the divisions of quality and action."
Two words carry the whole argument. Guna — quality, disposition, temperament. Karma — action, work, what you do.
Not janma. Not birth. The verse does not say birth.
The case built on it
Krishna states the origin of the fourfold order in a single line, and the criterion he names is quality and action — what a person is like and what a person does. Birth is simply not mentioned, in the one verse where it would have been most natural to mention it.
So the original design was a classification by aptitude, closer to a modern idea of vocation than to a hereditary caste. Everything cruel that followed came from later interpreters who replaced guna with janma. Fix the misreading and varna is not the problem — the corruption of varna is.
The three replies
Reply one: read the neighbouring verses
In chapter 18, the Gita returns to the four orders and assigns duties by qualities "born of one's own nature" — svabhāva-prabhavair guṇaiḥ. Critics argue that svabhāva, one's own inherent nature, was understood by essentially every classical commentator as something you arrive with, not something you choose. If your nature is given, and your duty follows your nature, the freedom in 4.13 narrows considerably.
Reply two: read what Arjuna is afraid of
In chapter 1, before any of the philosophy begins, Arjuna lists the disasters he fears from the war. Among them is varna-sankara — the mixing of varnas — which he presents as a catastrophe destroying families and ancestral rites.
That is the reply that lands hardest. Mixing is only a coherent fear if varnas are groups you are born into and can therefore be born between. You cannot dilute a category defined purely by aptitude.
Reply three: the verse that is quoted as inclusive
"Even those of sinful birth — women, vaishyas, and shudras — they too attain the supreme goal."
This verse is genuinely radical in one direction: spiritual liberation is declared open to everyone, in a world where many held that it was not.
But look at the sentence structure. Even they. The verse extends the door to three groups by naming them as the groups that would not otherwise be assumed to have it — and it does so by birth category, listing women alongside two varnas. The generosity and the hierarchy are in the same line.
Because the Gita is not a legal code and was never trying to be consistent about sociology. It is a philosophical dialogue set inside an epic, and it contains a line that clearly points away from birth and other lines that clearly assume it.
Which means both sides can quote it honestly. Side A quoting 4.13 is not lying. Side B quoting 1.41 and 18.41 is not lying either. They are quoting a text that does not speak with one voice.
The dishonesty only begins when either side pretends the other verses are not there. Which is, in practice, what almost every WhatsApp forward on this subject does.
It does not, in either direction. Hinduism has no single authoritative text, no council that fixed doctrine, and no equivalent of a canon law. It has a vast library composed over more than a thousand years in which almost every position can be sourced to something.
That is a genuine feature of the tradition, not a defect. But it means anyone claiming that "the scriptures say" anything definite about caste has chosen a shelf and called it the library.
9Four Ways To Read The Same Evidence
Every serious position on the origin of caste, laid out side by side, each written the way its best defender would write it.
All four of the positions below are held by intelligent people who have read the sources. They are working from the same hymns, the same excavations and the same genomes. Here is each one at full strength — followed by the objection it has the most trouble with.
The case: the Rig Veda mentions the four-varna scheme once, in its latest layer, and says nothing about heredity, marriage or purity. The Gita names quality and action, not birth. Families of mixed trades appear without comment. A servant's son becomes a Vedic teacher's student on the strength of his honesty. And the genetics shows widespread population mixing continuing for a long time — a society that had not yet closed. The rigidity is real but it is a later disease, not the original body.
Its hardest problem: the ranking in the Purusha Sukta is vertical and divinely sourced from the start, and the counter-examples are recorded as exceptional. A system with a top and a bottom, said to have been made by gods, is not merely a job list even on day one.
The case: mouth, arms, thighs, feet is not a neutral list. It places groups on a vertical axis and attributes that placement to a cosmic sacrifice, which puts it beyond human revision. The exceptions in the texts are recorded precisely because they were extraordinary, and one of them — Kavasha Ailusha — features explicit expulsion on grounds of birth. Arjuna's fear of varna-mixing shows that varnas were already understood as birth groups. And the endogamy lock closes long before any foreign arrival.
Its hardest problem: if birth-based rigidity was there from the start, the widespread population mixing visible in the genome for centuries afterwards is very hard to explain. Closed groups do not mix.
The case: Indian society shows real fluidity in the earliest sources and rigidity in the latest, and the periods in between contain repeated foreign rule. Colonial administration in particular counted, listed and fixed caste in ways that turned fluid local identities into permanent official categories — a process with a documented paper trail. Some version of this argument is accepted even by scholars who reject the strong form of it.
Its hardest problem: the timing does not work for the strong version. The genetic evidence puts the shift to endogamy at roughly two thousand years ago at the latest — centuries before the first Islamic rulers and well over a thousand years before the British. Outsiders can be shown to have hardened and formalised an existing system. They cannot have started it.
The case: the strongest version does not deny that hierarchy existed. It argues that a hierarchical society — of which there were many everywhere in the pre-modern world — has been retrospectively described in the language of unique atrocity, largely for political purposes; that colonial and missionary sources had motives to portray Hindu society at its worst; and that the four-varna scheme was a textual ideal that never governed most Indian lives, which were organised by region, occupation and local custom.
Its hardest problem: it is untestable at this depth of history and it is not really an argument about origins at all. It is an argument about the period we can actually document — court records, land records, temple records, travellers' accounts, colonial files. That evidence exists in enormous quantity. This position lives or dies in Part 6, not here.
Positions 1 and 3 are frequently held together, and the combination is politically comfortable: our system was fine, foreigners broke it. Positions 2 and 4 are almost never held together, because they contradict each other outright — one says the hierarchy was severe from the beginning, the other says it was never that severe.
Worth noticing when you see someone using both in the same argument.
10Nine Questions Nobody Asks
The blind spots. Things that almost never come up, on either side, and that change the shape of the argument once you see them.
Everything up to here has been the argument as it is normally conducted. This chapter is the argument as it almost never is. These are the questions I kept running into while reading, that nobody seemed to be asking out loud.
1. Why did India's hierarchy survive when everyone else's collapsed?
This is the question that should be at the centre and is almost never asked at all.
Hierarchy was universal in the ancient world. Egypt, Rome, China, Persia, Mesoamerica, medieval Europe — all of them had rigid, birth-based ranks, and several were far more violent about it than anything in the Indian texts. Roman slavery was industrial in scale. European serfdom bound millions to land they could not leave.
And all of those systems are gone. Not reformed — gone, as living social structures.
India's is not gone. It is weakened, illegal in many of its expressions, and under enormous pressure — but people still overwhelmingly marry inside it. So the interesting question is not "was India uniquely oppressive?" The honest answer to that is probably no. The question is: what made this particular system so much more durable than all the others?
And the likeliest answer is not cruelty. It is engineering. Most hierarchies are enforced from above — by a king, an army, a church. Remove the enforcer and the system decays. Caste is enforced horizontally, by thousands of separate communities each policing its own marriages. There is no head to cut off. That is why it outlasted every empire that ruled over it.
2. If varna began as work, doesn't that argue for change rather than against it?
Notice something odd about how Position 1 gets used.
The argument "varna was originally about work, not birth" is almost always deployed as a defence — of the tradition, and usually against reservation. But run the logic forward. If the system began as a flexible classification and later became a hereditary prison, then the hereditary part is the deviation. The reformer is the one being faithful to the original, and the traditionalist is defending the corruption.
People who make this argument are, without noticing, making the strongest available case that caste is a mistake that Hinduism itself should correct.
3. What would the record look like if the losers had written it?
Every text in this book was composed, memorised, transmitted and preserved by one narrow group: literate ritual specialists. For thousands of years, they were the archive.
That is not a conspiracy claim — it is just what literacy meant in the ancient world. But it has a consequence people skip. When we say "the texts do not mention widespread oppression," we are noting that the beneficiaries' own records do not mention it. That is exactly the source least likely to.
It cuts the other way too. Those same texts preserved the story of a slave woman's son composing hymns, and of a boy of unknown father accepted for his honesty. A pure propaganda archive would have deleted both. The record is biased. It is not fabricated.
4. Are we reading descriptions or complaints?
A rule is usually written down when it is being broken. "Do not do X" is evidence that people were doing X.
So when a later text angrily forbids intermarriage between varnas, there are two possible worlds behind it: one where the rule was obeyed and the text records the norm, and one where people were marrying across lines constantly and an alarmed authority was trying to stop them.
These are opposite societies producing an identical document. Almost every claim about how caste "actually worked" before the modern period runs into this wall, and very few people acknowledge it.
5. Why is there no memory of anyone deciding this?
Big social systems usually have a founding: a lawgiver, a conquest, a treaty, a council. Caste has none. No Indian text says "in this year, the following arrangement was established."
It appears to have assembled itself, over centuries, out of a very large number of local decisions by people who were not coordinating and probably did not think they were building anything.
That matters for the argument, because it removes the villain. It is far more comfortable to fight a system somebody imposed than one that grew. But it also removes an excuse: nobody imposed it, which means nobody outside is responsible for it.
6. What if the four varnas were never the real system?
Everyone argues about four categories. Almost no Indian has ever lived in four categories.
People live in jatis — thousands of them, tied to region, occupation and language, ranked differently in different districts, and frequently disputing their own placement in the four-fold scheme. Communities have argued for centuries about whether they are Kshatriya or Vaishya. The four-varna model was, for most of Indian history, a theoretical framework that local reality was pushed into afterwards.
Which means arguing about the Purusha Sukta may be arguing about a map rather than the territory. It matters enormously — because the map was used to justify the territory — but it is not the same thing.
7. Both sides need the Aryan question to come out a certain way. Why isn't that suspicious?
Consider how strange it is that a technical question about Bronze Age population movement has a predictable political answer attached to it.
If you can guess someone's view on migration from their view on reservation, then the evidence is not doing the work. And that applies in both directions — this is not a point about one camp. When a factual question sorts people by politics, it is being used rather than investigated.
8. Why did Buddhism reject caste and then leave?
Buddhism explicitly rejected birth-based worth. It spread across Asia and shaped half a continent. And it very largely disappeared from the land of its birth.
Whatever explains that is central to the story of why caste endured here — and it is a genuinely open historical question. This is Part 3's territory, but it belongs on the list because almost nobody arguing about caste origins brings it up at all.
9. Does knowing the origin actually change what we owe each other?
The last one, and the one that should be asked first.
Suppose we settled everything in this book tomorrow. Suppose it was proved beyond doubt that varna began as a gentle description of work and was hardened by outsiders eight hundred years ago.
What changes for the family that has been landless and excluded for twenty generations?
Nothing. Not one thing. Their position was produced by what was done, over centuries, not by what was originally meant. Origin stories assign blame. They do not undo consequences.
11What The Evidence Can And Cannot Settle
The honest scorecard. What we actually know, what we merely believe, and what nobody can answer at all.
After all of that, here is the part almost no book on this subject will give you: a clear statement of the limits.
| Question | Status | Best current answer |
|---|---|---|
| Did population mixing stop and endogamy begin in ancient India? | Settled | Yes — roughly 1,900–4,200 years ago |
| Did the Indus Valley have caste? | Unknowable | Script undeciphered. No basis for any claim. |
| Did Steppe-related people move into South Asia? | Strong consensus | Yes, second millennium BCE |
| Did those migrants create the caste system? | Not supported | Ancestry is spread across all levels of the caste order |
| Does the Rig Veda describe a ranking? | Yes | Vertical, divinely sourced, stated once |
| Does the Rig Veda make varna hereditary? | No | Heredity is not stated in the text |
| Was there movement between varnas? | Yes, some | Recorded — and recorded as remarkable |
| When exactly did varna become jati? | Open | Gradual, over centuries — see Part 3 |
| Was historical caste oppression exaggerated? | Answerable | But not from this evidence — see Part 6 |
The three things nobody can tell you
- What ordinary people thought. Every text we have was composed and preserved by a literate priestly minority. We have the ideology of the people who wrote. We do not have the view from the village, because the village did not write.
- Whether the texts described society or prescribed it. This is the deepest problem in the whole field. A text saying "a Brahmin should do X" might be reporting what Brahmins did, or complaining that they were not doing it. Legal and religious codes tend to be written when a rule is being broken. We often cannot tell which we are reading.
- Why the lock closed. Genetics gives us the date with real confidence. It gives us the reason with none at all. Every explanation you will read for why India became endogamous — priestly power, land, agriculture, conquest, ritual purity — is a hypothesis, not a finding.
Go back to the table in Chapter 1 — the four possible histories and what each implies for reservation today.
Here is what struck me hardest while writing this part: almost nothing in it changes the answer.
Suppose Position 1 is completely correct and varna began as a flexible description of work. It still hardened. It was still closed for something like two thousand years. A person born into a group excluded for eighty generations is not less excluded because the exclusion started as a metaphor.
Suppose Position 3 is correct and foreigners did the hardening. The hardening still happened, and the people who inherited its effects are the same people.
The origin story determines who we blame. It barely touches what was actually done, to whom, and for how long — and it is that which any remedy has to answer to.
Which is why the argument you are about to read in Parts 3 and 6 matters far more than the one in this book, and gets far less attention. Origins are more fun to argue about than evidence.
12The Five-Line Nichod
Everything in this book, compressed. If you remember nothing else, remember these.
PART 2 — BEFORE CASTE
- Genetics gives us a date, not a reason. Population mixing across India was widespread until roughly 1,900–4,200 years ago, and then it stopped. That is when the lock of endogamy closed — long before any Islamic or British arrival, and long after the earliest Vedic period.
- The Indus Valley cannot answer this question. No palaces, no royal tombs, no monuments to rulers — and no readable script. Anyone claiming the Indus proves either equality or caste is filling silence with preference.
- The four-varna hymn appears once, late, and says nothing about birth. The word shudra occurs exactly once in the whole Rig Veda, in Mandala 10. But the arrangement it describes — mouth, arms, thighs, feet — is unmistakably vertical and is credited to the gods.
- Both sides can quote scripture honestly. The Gita says quality and action in 4.13, and assumes birth groups in 1.41 and 18.41. Vishvamitra crossed varna; Kavasha Ailusha was expelled for his birth. The tradition does not speak with one voice, and pretending it does is the only real dishonesty available.
- The origin barely changes the verdict. Whether varna began as work or as rank, whether foreigners hardened it or Indians did, it closed and it stayed closed for roughly two thousand years. Origins tell you whom to blame. They do not tell you what was done — and a remedy answers to what was done.
Coming next
Four categories became three thousand communities. A metaphor became a wall. Part 3 is the mechanism: the Dharmashastras and the actual verses of the Manusmriti — including the ugliest ones, quoted in full and with the honest debate about how far they were ever enforced. Endogamy as the single lock that froze everything. Where untouchability came from, including Ambedkar's own theory and its competitors. And Buddhism and Jainism as the first great revolts against all of it.
This is the part where the argument stops being about metaphors and starts being about rules with punishments attached.
A closing thought: if you finished this part feeling that your side did not win cleanly, that is the correct reaction, and the evidence is responsible for it rather than me. The clean win is available only to people who stop reading early. — L.S.
Sources & further reading — Part 2
Every factual claim in Part 2, and where it comes from.
Genetics
Shinde, Narasimhan et al., "An Ancient Harappan Genome Lacks Ancestry from Steppe Pastoralists or Iranian Farmers," Cell, 2019 — the Rakhigarhi individual (I6113); Iranian-related and AASI ancestry, no Steppe component. · Narasimhan et al., "The Formation of Human Populations in South and Central Asia," Science, 2019 — ancient DNA from 523 individuals; Steppe-related ancestry entering South Asia after the mature Indus period; Indus-periphery outliers at Shahr-i-Sokhta and Gonur Tepe. · Moorjani et al., "Genetic Evidence for Recent Population Mixture in India," American Journal of Human Genetics, 2013 — genome-wide data from 73 groups; ANI–ASI mixture dated to approximately 1,900–4,200 years ago; essentially no mixture before c. 4,000 years ago; widespread mixture thereafter, followed by the shift to endogamy.
Archaeology
Green, "Killing the Priest-King: Addressing Egalitarianism in the Indus Civilization," Journal of Archaeological Research, 2020 — the absence of evidence for a managerial ruling class, and the case that social complexity does not require one. · Standard summaries of Indus urban planning, standardised brick ratios and weights, the Great Bath, and Meluhha references in Mesopotamian records. · The counter-position on limited burial samples and differential access is drawn from the same literature, which contains both readings.
Texts
Rig Veda 10.90.11–12 (Purusha Sukta) — transliteration and translation as given. The word śūdra occurs in the Rig Veda only in this hymn. Mandala 10 is generally placed in the latest stratum of the collection on linguistic and metrical grounds. · Rig Veda 9.112.3 — the bard, the physician and the grinder of corn. · Chandogya Upanishad 4.4 — Satyakama Jabala. · Aitareya Brahmana — Kavasha Ailusha, son of a dasi. · Bhagavad Gita 4.13 (guṇa-karma-vibhāgaśaḥ), 1.41 (varṇa-saṅkara), 18.41–44 (svabhāva-prabhavair guṇaiḥ), 9.32.
On the Aryan question
The migration model is the mainstream position among geneticists, archaeologists and linguists publishing peer-reviewed work as of 2026; Out of India remains a minority position outside that consensus. The Mitanni treaty invoking Mitra, Varuna, Indra and the Nasatyas, and the Kikkuli horse-training text using Indo-Aryan numerals, are standard evidence in the linguistic case. The absence of archaeological evidence for a violent conquest of the Indus cities, and the role of climate and river-system change in their decline, are also mainstream findings — which is why the older Invasion Theory is treated in this book as superseded.
A note on what is missing — and on corrections
Part 2 makes no claim about Manusmriti, untouchability, the medieval period, colonial rule, or whether historical caste oppression was severe. Those belong to Parts 3, 4, 5 and 6, where the evidence is documentary rather than inferential and can be tested properly. Anything marked CONTESTED CLAIM here stays contested until evidence is produced — including claims made by people I might agree with.
Sanskrit transliteration and dating in this part follow standard scholarly conventions, but this is a field where specialists disagree in good faith, and where new ancient-DNA results have repeatedly revised the picture within the last decade. If you find an error, I want to know. Corrections will be carried in later parts.