In the near future the ancient DNA group led by David Reich will publish a bunch of stuff, and one paper will note that the variation in steppe ancestry in the Bronze Age Greeks did not have class implications. In other words, the ancient Greeks did not have a caste system as you might find in India, despite the way the Spartans treated the helots or Messinians. Of course, the Indo-Europeans did have a ‘tripartite’ caste system of rulers and warriors, priests and commoners. In the Indian varna system, this was translated into Kshatriyas, Brahmins and Vaishyas. But it is found elsewhere, including among the German Saxons. But to my knowledge nothing like the Indian caste system has been found genetically in these ancient populations; some individuals have more “farmer” ancestry in initial generations, but this is all smoothed away by admixture.
India is different. It has jati-varna, with varna being caste as you would understand, but jati being one of thousands of endogamous “communities” in the subcontinent. At first, like many, I assumed this had something to do with the Indo-Aryan migration into the subcontinent, but I noticed a few things early on. First, there are apparent genetic differences between groups in Non-Aryan South India that are not Brahmin. For example, between Nadars and Dalits. These differences correlated to global biographic fraction differences. Dalits have more AASI (Ancient Ancestral South Indian). Mind you, I believe South Indian “Dravidians” were strongly shaped by the Indo-Aryans, so that’s not dispositive. Nevertheless, Non-Aryan cultural regions have this institution, and, jati is peculiarly India, even if varna is not.
But, what has shifted my view is looking at admixture variation in “Indus Valley Periphery” samples in samples dated from 3000-2000 BC. There’s a wide range of AASI. Why? Well, admixture in structured populations takes time. But there’s something suspicious to me about this variation combined with India’s later endogamy and the mystery of how the Indus Valley Civilization organized itself. There seems to have been very little stratification in a way we would understand it from Egypt or Mesopatamia. Were they anarchists? I doubt it. The early emergence of jati may explain the IVC sociopolitical system. The Indo-Aryans, when they arrived, were simply integrated into the framework.
Ancient DNA will prove me right or wrong. But I’m putting my cards on the table.
The easiest explanation I can think of for Jati stratification comes the nucleus of India’s settlement patterns – high density and “clean areas”. Even rural villages can have thousands of residents.
India also has a unique class structure in that you have different levels of “cleanliness” that define your position, with the observance of different rites and life regimens. Polynesians did have a similar caste system, but without the genetic differentiation.
Basically agriculturalists found the villages, towns and cities. As they expand they run into hunter-gatherer populations. These populations can either be displaced, or enter the periphery in a symbiotic relationship. You’ll have some intermarriage, but it will be in “layers”.
The roles that these newcomers find, though, will be less desirable and “unclean”. In some cases, they will be coerced into them. If you look at a Tamil village, you literally see a high caste nucleus with lower castes and dalits on the periphery
These communities are also forced to use downstream water, and in some cases, must go around the village nucleus instead of through it.
You can seen this dynamic with the Irulas and Panniyas. Until recently they might have been considered tribes, but they’re forced into the agriculturalist world. The Irulas are basically pest controllers, and their handling of dead rats and snakes puts them in a “unclean” and peripheral margin, while the Panniya were basically exploited Labour through debt.
The Indo-Aryans seem to have spread through their religious rites and language, so they easily mix with the priestly and “clean” groups while keeping the hierarchy.
Interesting.
Is Indra a BMAC god?
“what has shifted my view is looking at admixture variation in “Indus Valley Periphery” samples in samples dated from 3000-2000 BC. There’s a wide range of AASI. Why? Well, admixture in structured populations takes time. But there’s something suspicious to me about this variation combined with India’s later endogamy and the mystery of how the Indus Valley Civilization organized itself. There seems to have been very little stratification in a way we would understand it from Egypt or Mesopatamia. Were they anarchists? I doubt it. The early emergence of jati may explain the IVC sociopolitical system. The Indo-Aryans, when they arrived, were simply integrated into the framework.”
The South Indian Neolithic dates to about 2500 BCE. Before that people in areas that are now Dravidian were basically hunter-gatherers. It makes sense for early admixture to coincide in time with the emergence of agriculture in the South for for AASI to vary most beyond clinal variation at the dawn of this transition.
Harappan and Dravidian were probably two completely distinct languages from different language families. So, Harappan antecedents to caste could go back a long way, but it seems like the wholesale cultural assimilation of Dravidian people’s into a larger Hindu religion and South Asian culture didn’t happen until a first wave of Indo-Aryan conquest that conquered almost all of Dravidian civilization. South India’s LD date of most recent admixture with Indo-Aryans is older than in the North of India where there appears to have been a subsequent second round of Indo-Aryan admixture.
Thus a sequence of descendants of Harappan civilization integrating in Indo-Aryans into their system first, followed by conquest of Southern India by the fused culture, followed by a separate and distinct integration of Southern Indians into the fused Harappan-Indo-Aryan culture makes sense and would explain different patterns in the South than the North genetically in India. Also, it could be that fused Indo-Aryan expansion was led by integrated Harappans with less steppe ancestry than Indo-Aryan leaders in the North in former Harappan territory, since they had more prospects for fame and fortune in newly conquered areas than in the places they grew up were they were “second rate” by comparison to more “pure” Indo-Aryans.
The disconnection in the linguistic diversity of Dravidian (which is too slight to originate in 2500 BCE with the South Indian Neolithic when Dravidian would plausibly arise) can be explained as being due to reconquest of most former Dravidian territorial from a small remaining nucleus of linguistic Dravidian territory not conquered by Indo-Aryans with that dialect being the source language for all of “rebooted” Dravidian languages.
Genetics suggest that jati weren’t as profoundly endogamous as they are now until the mid-first millenium CE, after the collapse of IVC and after the Indo-Aryan arrival. It is hard to know if they were there before Indo-Aryans arrived and they were just less strictly endogamous, or if they were a new invention driven by 1st millenium CE forces absent in other Indo-European conquered regions. (I agree that jati is a cultural and political feature that is not Indo-European in origin.) One strong argument for them arising in the first place in something like the present form only when jati endogamy is clearly visible genetically is the powerful influence of founding populations from that time frame that were much smaller than current jati populations. This shouldn’t be as strong if they existed much longer but were only moderately endogamous before then. It could be that they were necessitated by a weak political state in that era which was tumultuous enough to also lead to the establishment of new religions including Buddhism with people looking for major lifestyle changes to address discontent with the status quo way of living at that time yet lacking engineering type technology to deal with it so they turned to social engineering.
Rig vedic mentions of jati also come very late in these texts, further arguing against a pre-Indo-Aryan origin for the practice.
I’m not convinced that IVC wasn’t “stratified” at all. Big cities by the standards of that era almost necessarily imply some stratification, but certainly it appeared to be much more political and culturally united than the Nile Valley or Mesopotamia, almost from the outset until the verge of its collapse, and it appears to have been fairly urban fairly early. At a minimum, its city-states seem to be part of a single federal state or a single empire. The lack of large scale warfare prep and consciousness until the very late Harappan is notable. But the stratification need not have been hereditary in a “hard” sense.
Jati imply economic specialization of fairly complex society, and implies that these economic conditions were fairly stable for very prolonged periods of time (centuries). You need a strong enough state and civil society to allow complexity in society to persist, but not such a strong state that it is able to “cut out the middleman” and interface directly with the individual people without civil institutions mediating that relationship. Some aspects of jati in the uniquely Indian context also imply a legacy of responses to one or more pandemics since it is focused on being “clean” in a way that non-Indo-European society caste does not.
One might see jati as having some parallels to religious taboos like kosher and hallal religious laws and other religious dietary restrictions in the Eastern Mediterranean and Africa, that play out differently for some reason.
I’m not offering answers, but I am collecting facts. And, I’m pretty skeptical of the pre-Indo-Aryan theory of jati even though I don’t think it was Proto-Indo-European in origins either.
That is wild to think that not even the Sons of Indra could break the caste system, just accommodate themselves into it. The Indian version of the Mongols getting absorbed into Imperial China despite their best efforts.
The Indus_Periphery samples are basically a bunch of foreigners (possibly with some local admixture, possibly without), that we find in the BMAC city state of Gonur and the Shah-i-Shokta (“burned city”) site in Iran. At the most extreme, some of the samples separated by 1000 years of time.
Quick plot of samples that made it to David’s G25: https://i.imgur.com/1XqOT7e.png
So I think there’s a question of whether they all had to come from one place, where social stratification was in place to maintain genetic variation, or whether they just came from multiple sites with structured variation between site.
I don’t know if there was much variation between grave goods between different samples with varying levels of AASI that indicated social hierarchy.
Of the samples that made it onto Davidski’s Global_25, both the samples noted to have poor grave goods were I8726, who is the most West Eurasian of the samples that made it over (dated earliest at approximately 3000 BCE), and I8728, who was dated 500 years later and is the most AASI.
The sample I11459 who was noted to have “very high proportions of ancestry related to AHG and a H1 Y chromosome haplogroup typical of South Asia” seems to me to the most rich grave out of the Indus_Periphery samples at Shah-i-Shokta, from the supplement (https://reich.hms.harvard.edu/sites/reich.hms.harvard.edu/files/inline-
files/2019_Science_NarasimhanPatterson_CentralSouthAsia_Supplement.pdf), although that’s subjective.
At Gonur the number of samples is much less and the variation in AASI too. We find that sample I11041 who was in a wagon burial (or burial with wagon at least) in the Royal Necropolis did not have much AASI ancestry, but there isn’t much variation in AASI in the Indus_Periphery samples found at Gonur.
I think you’re right here that there is an important question of why the samples that we find in Indus_Periphery have this varying AASI. But it’s hard for me to agree confidently say that its from social structure within Indus_Periphery sites, rather than the foreigners being migrants from different places in IVC (perhaps in some cases with some local ancestry too).
Furthermore, a couple more counterarguments:
1) The caste structure we see in India today is theorized by Reich to mostly reflects a homogenized AASI+Iran_N mix – called ASI – with a homogenized Steppe_MLBA+Iran_N+AASI mix – called ANI. This cline of mixed populations in theory post-dates the Indo-Aryans.
This means that even if the ideas of caste predated the Indo-Aryans, the actual existing groups themselves, in theory formed after that. If the Reich model is correct. (So in a sense caste would have had to die and be reborn under this scenario).
E.g. looking at the figure from Narasimhan’s paper that you posted, the Indian cline is a single cline and its not from steppe ancestors mixing into the different points on the Indus_Periphery cline.
2) The samples at Swat_IA from the early Iron Age (approx around 1100-800 BCE), with steppe ancestry, were checked by Harald Ringbauer’s methods for signs of recent endogamy and I believe nothing was found indicative of either cousin marriage or endogamy, so like ohwilleke says, not a clear indication of any endogamous community.
Further I think the samples at Swat from multiple sites, although there’s variation, had a similar average apart from maybe one site reliant on three samples (Aligrama_IA). The variation in Swat_IA seems to be driven by a large main cluster with a similar average, and then some outliers at the large Katelai and Loebanr sites, and a few samples at the Aligrama site that have high AASI. Again the question; do the AASI rich outliers have rich/poor graves?
Narasimhan notes on the Aligrama site: “Three individuals are assigned the Aligrama2_IA analysis label (which is the same as their split label). Although the three individuals are possibly from three of the 35 protohistoric graves of Aligrama excavated in autumn 1981, these individuals are not grouped with the individuals from the other Swat Protohistoric Grave sites that are assigned the SPGT analysis label, because we empirically observe that they have less Steppe pastoralist-related ancestry than the SPGT. The archeological metadata related to these samples is uncertain. The provenance of the samples is instead reconstructed on archeological grounds”. Because of this lack of clarity, no information is given about the grave goods other than “No further archeological information is available at the present time. A study of the graveyard is in progress by the excavators of the site.”
Graphically, the Swat_IA on the cline: https://i.imgur.com/HFZjqtn.png
So all in all, I’m not convinced that Indra the R1a-an (or one of his later scions) is exonerated in the matter of caste! Some ideas around marriage networks that maintained genetic variation in AASI before the Indo-Aryans may have been possible, but those varying groups largely don’t seem ancestral to the varying groups that exist today, and there’s no evidence that they were hierarchal that I can see, even when we get actual steppe admixture.
I guess summing up the above, I think we could define a little more what we’re saying:
That structured AASI proportions and associated endogamy within sites, and associated with social status, was present in IVC and steppe groups admixed with these structured populations without disturbing relationship of AASI with social status, nor long term endogamy?
Possible but the I_P samples aren’t good to test that and moreover neither they nor the Swat seem to me to show any apparent clear association between AASI and social status in burials. Neither does, I think, IVC hold much indication of social status difference in burials to test this, even if we could get the adna?
Moreover, the standard model suggests present-day jati populations arose most parsimoniously from ANI and ASI, not from steppe and a variety of populations on the I_P cline.
If we’re saying a different thing though, which is simply that a structured set of marriage networks that prevented panmixia was present in IVC – potentially without hierarchy – and influenced jati/caste, that seems like I’d give it a higher probability. But direct adna is still crucial to test due to inherent limits of the I_P set.
Btw, Razib apparently Niraj Rai revealed some new ancient dna findings from India at (of all things!) a literary festival. (Hat tip anthrogenica)
This is the link: https://www.youtube.com/watch?v=FevPClW3Lm8
Following are what it seems to show, bits in quotes are an anthrogenica user’s words, not Rai’s. I have yet to go through the full presentation to confirm all but just skimmed it so far:
– “Kashmir – no steppe dna circa 1500 BCE, but 200 AD steppe rich”. Possibly this is more evidence against the idea that steppe_MLBA to South Asia significantly pre-dated the Swat_IA generally samples around 1000 BCE, or maybe not if Kashir is off the beaten track.
– South Indian Megalithic Tamil Nadu samples from 600-300 BCE – show no apparent steppe ancestry. This might fit with a relatively late integration of South India into the population with the North, possibly consistent with the LD patterns ohwilleke is talking about.
– “Naga hills 3000-4000 YBP – unlike at present no ‘Han’ ancestry seen, but like at present ancestry connection with Burma and Tibet”. Not sure what this one means, the samples will be approx 1000 BCE -https://www.researchgate.net/publication/352465037_Ancient_DNA_investigation_at_Photangkhun_Longkhap_Nagaland_India.
It sounds like the Oakaie sample from Myanmar at ~1000 BCE which was clearly mostly Sino-Tibetan in ancestry, but I have no idea what no ‘Han’ ancestry would mean, since Tibetans are Northern Han like… Some subtlety in layers of migration from China? Maybe related to the spread of Dai like ancestry?
– Some stuff around Ahom people but I can’t really tell what the presentation is showing.
– Sanauli burial at 2000-1900 BCE, not confident about ancestry (low quality sample) but sure that lacks steppe related ancestry. I’m not sure about the debate about whether this is a chariot or not, but if horse drawn, if suggests that domestic horses of the Sintashta type and the chariot as a technology were moving faster than the movement of steppe ancestry, which is what we see in the Near East anyway.
– Some stuff about migration to Bali by Indian settlers from South India by 300 BCE. This might match with the South Indian like genetic character of migration into Cambodia that has been detected.
Finally some exciting things from India itself are firmly on the horizon…
“Some stuff about migration to Bali by Indian settlers from South India by 300 BCE. This might match with the South Indian like genetic character of migration into Cambodia that has been detected.”
A search for population structure and endogamy and founder effect concentrations of deleterious genes in the South Asian diaspora ca. 300 BCE, which is before genetic estimates of the commencement of endogamous jati in South Asia (ca. 500 CE +/- a few hundred years), would be a good way to partially test the hypothesis that jati are pre-Indo-Aryan. The genetic data point to jati formation during the Classical Period of South Asian history. https://en.wikipedia.org/wiki/History_of_India#Classical_period_(c._320_%E2%80%93_650_CE) with the Gupta empire looking particularly attractive.
It would also be interesting to see if there is systemic regional variation in the onset of jati endogamy estimated genetically by region.
Another point worth mentioning is that the ancestry proportions of different Varna in India as well as the sex differentiation of uniparental ancestry in India strongly indicates that in the formative period when ANI was arising from a mix of Indo-Aryan migrants (mostly male) and indigenous North Indians (mostly female at the first generation of admixture) and in many subsequent generations, that Varna were not strictly endogamous for a long time. The level of population structure in this formative era is more akin to social class based assortative marriage in Europe and West Asia than to a true tight caste divide as we understand it today.
The genetic fingerprints of cross-Varna marriage are all over South Asian population genetics. And, we know from numerous studies that late Copper Age and Bronze Age Indo-Europeans, pretty much everywhere, were patrilocal and married wives from distant places or at least outside their own clans, exogamously.
So, the notion of endogamous structure arising from Indo-Aryan migrants just doesn’t really ring true, either from the first contact era between IVC and Indo-Aryan, or before it (with the analysis from @Matt on that score ringing true in that regard).
It was be nice to see some Stronium analysis of IVC era human and animal remains to get a handle on how much geographic mobility there was between IVC cities.
Also, FWIW and not precisely on point, I’d really like to get better data on Y-DNA T in South Asia with more detailed haplotype data and analysis of carriers of it. I think that there is an interesting story lurking there.
@Matt
>Neither does, I think, IVC hold much indication of social status difference in burials to test this, even if we could get the adna?
As in difference in how well made a grave is and how many goods were deposited in it?
Or do you mean correlation of that with adna? Because in that case, there is only one sample from the IVC proper zone. So a comparison cannot be made yet.
@DaThang, yeah, exactly I mean both. To work out whether we had social status correlations related differences in ancestry in the IVC we’d need
1) burials showing social status differentiation – better made graves, and more grave goods – which supposedly I think IVC lacks (although I’m not sure on this and I’m happy to be corrected – this may be a meme I’ve absorbed and not accurate)
2) dna/strongly correlated anthropometry from IVC itself to show whether these burials had differences in ancestry.
Which as you say, we don’t have yet for a comparison to be made (if ever, if there is a lack of status differentiation in burials in the first place. but like I say, I only really have a meme level understanding of if this is the case).
@Matt
1-Some graves are better made with more grave goods. Better made graves can be put in 2 others one are the brick lined graves which only have females. The other group, not brick lined but nonetheless with more goods than normal graves can have either genders. This is sort of what you’d expect if eliteness, instead of being concentrated in a few bloodlines like pharaohs is instead diluted to a large number of people that make up a non-trivial minority elite percentage(s) of society.
2- We don’t know if this corresponds to ancestral proportions in IVC proper since the sample size is 1 for now. Personally in the absence of a more recent migration deeper into the eastern zone of the ivc in the calcholithic from Iran, the mixture would have happened long before even the early harappan period began.
Still you might get heterogenous results, like western ivc with more recent chalcolithic admixture having class correlate it, while the eastern zone may not have the correlation. Or if it does then somehow it maintained it all the way from the Neolithic to the ivc proper period.
@ohwilleke; “The level of population structure in this formative era is more akin to social class based assortative marriage in Europe and West Asia than to a true tight caste divide as we understand it today.
The genetic fingerprints of cross-Varna marriage are all over South Asian population genetics. And, we know from numerous studies that late Copper Age and Bronze Age Indo-Europeans, pretty much everywhere, were patrilocal and married wives from distant places or at least outside their own clans, exogamously.
So, the notion of endogamous structure arising from Indo-Aryan migrants just doesn’t really ring true, either from the first contact era between IVC and Indo-Aryan, or before it (with the analysis from @Matt on that score ringing true in that regard).”
Don’t have time for a fuller comment, but this seems likely to me. Though would qualify that there must be some kind of structure that maintained the Brahmin divide we see where they have not only a higher West Eurasian related level of ancestry – which could be explained under the model of only ANI and ASI populations forming and then admixing – but they have a higher ratio of Steppe:AASI/Iran_N related ancestry for their level of West Eurasian. So that must have been explained by something…
Question whether it relates to marriage patterns within culture, or because the Brahmin groups tend to descend from some particular regional population that formed in this horizon? And non-Brahmin groups like this, such as Rors. And the curious historical era sample I6893 R1a-Z93 from Swat from 200 BCE, where he has an excessive ratio of steppe relative to her overall West Eurasian shift, more so than Brahmin people today it seems to me. (Less West Eurasian shifted than the Loebanr outlier I12138 woman from the early Iron Age, but more steppe shifted as a share of his ancestry.) (most of the other historical era samples seem broadly within the main ANI:ASI cline).
“Though would qualify that there must be some kind of structure that maintained the Brahmin divide we see where they have not only a higher West Eurasian related level of ancestry – which could be explained under the model of only ANI and ASI populations forming and then admixing – but they have a higher ratio of Steppe:AASI/Iran_N related ancestry for their level of West Eurasian.”
There is a divide but not nearly as strong that the divide for jati in the last 1500 years.
Children of several of my cousins will not cluster with KP samples …
Looks like I made a typo in the first part of my comment. What I meant was the richer graves can be put into 2 groups:brick lined and non-brick lined. Brick lined were all females while non-brick lined were mostly males as far as I remember.