Mughal Emperor Akbar
In Strange Parallels Victor Lieberman made a reference to “Turkicized Pathans.” The very term has been gnawing at me. To get some sense of the context, Lieberman was sketching out the impact of Islamic civilization upon Indian civilization. Sometimes this “impact” was very literal. The Arab armies had rolled into Sindh in the 8th century, but that influence upon India was militarily marginal. The first real Muslim raider of consequence was Mahmud of Ghazni, a Turkic raider from what is today Afghanistan, who famously plundered the palaces and temples of North India circa ~1000. But even here the the impact is arguably superficial. Mahmud of Ghazni’s raids did not lead to a large Indian domain under his direct rule except in Punjab. Rather, these sallies into India were sources of supplementation to his broader fiscal resources. He was still fundamentally a Central Asian potentate fixated on Central Asian concerns. The real rise of Islamic civilization in India was precipitated by the Delhi Sultanate, a series of short-lived polities beginning circa ~1200 which dominated the Indian subcontinent for centuries, until they were superseded by the far more robust Mughal Empire.
These Indo-Islamic dominions were often dominated by individuals of Turkic identity. By this, I mean that they were from a lineage of Turkic tribes which had filtered into the world of Islam in the centuries before 1000, enslaved or enrolled in the armies of Muslim warlords. But eventually these pawns turned the tables on their erstwhile masters and snatched the keys to the kingdom for themselves. Mahmud of Ghazni’s own family were originally servitors of the Iranian Muslim Samanid dynasty. But just as Rome was enslaved by Greece culturally after its conquest of Hellas, so many Turks freely granted the manifest superiority of the Persian language in the domain of culture. Therefore the irony is that the Persian language spread as the elite cultural vehicle along with the expansion of the Turks west and east, culminating with the rise of the Ottomans and Mughals. Therefore you had a situation in Mughal India where the ruling dynasty, which was of proud Turco-Mongol origin along the paternal lineage, patronized Persian was the language of the court and administration more generally.
But what about the Afghans? They were not invisible. Along with the Turks and Persians, who came with the sword and quill respectively to serve in the courts of India’s Islamic rulers, came auxiliaries of Afghans, mostly Pashtuns. Though a majority of the dynasts seem to claim Turkic antecedents, some are self-consciously Afghan. For example the Lodi dynasty. The influence of these people is evident in India today insofar as upper class Muslims often refer to themselves as “Pathans,” presumably pointing to an origin outside of Indian proper.
To me it seems that the development of the Pathan over time is hampered by the fact that they are a people who did not develop their own independent robust high culture. A variety of Persian was the language of high culture. The Other par excellence was the Turk. The Pathan was a background figure, the illiterate peasant or nomad which was of no concern, and who integrated themselves into a Turkic or Persian world and identity when they rose above their station. In this way I wonder if they resemble Kurds, another highland Iranian people who have persisted over the centuries but seem strangely invisible as great empires rise and fall. This invisibility, a decentralized lack of reliance on elite institutional structures, may be one reason that a coherent Kurdish or Pathan identity persisted over so long, and continues down to this day, despite the spread of Turkic language in many zones of their broader region.
Though I probably know more about Indian history than most of you I’m still somewhat in the dark as to the detailed relationship of between Turks and Afghans in Afghanistan and India during this period. Much of the literature focuses on the faction between immigrants from outside India vs. those who were native-born, or between Muslim elites and Hindu elites. Secondarily there were divisions between Shia and Sunni, with Persians looming large.
I don’t have fluency in the languages to do primary research, but I do have genetic data sets to play with. I pruned my populations down to a few which were designed to explore the nature of East Asian ancestry among other groups, in particular Pathans. The population set has ~90,000 markers, and I ran ADMIXTURE across many K’s, with 9 seeming to be the most illuminating for my purposes. Below are three bar plots, one which shows population averages, and two which focus on specific populations on the grain of individuals. There are also two plots which visualize genetic distances between the hypothetical ancestral populations, labeled by modal group.
[zenphotopress album=288 sort=sort_order number=5]
When you analyze East Asian data sets they always tend to divide first into a northeastern and southeastern component. In this case I have a northern Turkic group, Yakuts, in my population set. Since the data set is “Pakistan-centered” you see divisions of ancestral groups with a focus on that region. If I overloaded with Western European populations the outcomes might be very different in absolute terms. As per what other genome bloggers have found using a Pakistan-centered data set South Asian populations can be separable as admixtures of three broad elements:
1) A north Pakistan centered element, modal in the Burusho ethno-linguistic isolate. This element seems rather distant from other West Eurasian components, and is broadly correlated with “West Asian” in other runs (though the overlap is imperfect because of the Pakistani bias of this data set).
2) A south Pakistan centered element, modal in the Brahui, a Dravidian ethno-linguistic group surrounded by the Iranian speaking Balochi, with whom they share most cultural features except language.
3) A more South Asian general element, which is represented here by the Gujarati_B sample (probably Patels). It’s position on the Fst MDS is pretty much where you would expect from a South Asian population which is an admixture of West Eurasian and non-West Eurasian populations. Interestingly the proportion of this element in the Balochi and Brahui is in the same neighborhood as among the Cambodians. On the face of it I’m skeptical that there was a mass migration of South Asians to Cambodia, despite the Indic associations of early Khmer society. Rather, it seems more likely to be evidence of an ancient South Eurasian substratum which spanned the Indian subcontinent and Southeast Asia. On the other hand, Malays and Cambodians exhibit evidence of South Asian ancestry even when the Andaman Islander component is extracted out. Looking at other groups I’m still strongly leaning toward the assumption that this is an artifact though looking at some other results at the linked plot. But it will be something to investigate more closely in the future.
But there are other components at low proportions among the Pakistanis aside from the “big three.” Here are the population breakdowns in tabular form:
| Population | Dai | Yakut | Burusho | Baltic | Brahui | Yemeni | Bantu | Sardinian | Gujarati |
| Brahui | 1% | 1% | 8% | 3% | 71% | 6% | 3% | 3% | 6% |
| Balochi | 1% | 1% | 12% | 3% | 62% | 6% | 2% | 2% | 10% |
| Makrani | 0% | 0% | 12% | 1% | 61% | 9% | 6% | 4% | 6% |
| Sindhi | 2% | 1% | 21% | 4% | 32% | 2% | 4% | 2% | 32% |
| Pathan | 2% | 3% | 27% | 11% | 22% | 5% | 1% | 2% | 27% |
| Iranians | 1% | 1% | 30% | 4% | 15% | 30% | 3% | 14% | 3% |
| Uzbeks | 14% | 25% | 18% | 18% | 7% | 8% | 0% | 4% | 5% |
| Turks | 1% | 4% | 25% | 12% | 7% | 27% | 0% | 22% | 1% |
| Syrians | 1% | 0% | 20% | 3% | 5% | 44% | 4% | 23% | 1% |
| Hazara | 21% | 29% | 27% | 7% | 5% | 4% | 0% | 2% | 5% |
| Gujarati_b | 1% | 1% | 2% | 2% | 3% | 2% | 0% | 2% | 88% |
| Georgians | 0% | 0% | 43% | 5% | 1% | 24% | 0% | 27% | 0% |
| Chuvashs | 2% | 20% | 3% | 71% | 1% | 1% | 0% | 1% | 1% |
| Cambodians | 86% | 1% | 0% | 0% | 1% | 0% | 1% | 0% | 10% |
| Belorussian | 0% | 0% | 1% | 84% | 1% | 2% | 0% | 11% | 1% |
| Burusho | 8% | 5% | 55% | 2% | 1% | 0% | 1% | 0% | 28% |
| Lithuanians | 0% | 0% | 0% | 95% | 0% | 1% | 0% | 4% | 0% |
| Yakut | 1% | 94% | 0% | 4% | 0% | 0% | 0% | 1% | 0% |
| Han | 83% | 17% | 0% | 0% | 0% | 0% | 0% | 0% | 0% |
| Yemen Jews | 0% | 0% | 0% | 0% | 0% | 99% | 0% | 1% | 0% |
| Dai | 100% | 0% | 0% | 0% | 0% | 0% | 0% | 0% | 0% |
| Sardinian | 0% | 0% | 0% | 1% | 0% | 2% | 0% | 97% | 0% |
| Miaozu | 89% | 11% | 0% | 0% | 0% | 0% | 0% | 0% | 0% |
| Bantu Kenya | 0% | 0% | 0% | 0% | 0% | 2% | 98% | 0% | 0% |
I’ve highlighted Pakistani populations, and bolded the modal fraction for all groups. The important point is to look at the ratio between “Yakut” and “Dai.” This should be an indication of the “Turkicness” of a population, with higher ratios implying more Turkic ancestry. The Burusho are a good test case. They exhibit little intra-population variation in the non-trivial East Asian proportion. But they’re somewhat biased toward a Dai component. I suspect that this balance is evidence of Tibetan admixture, as opposed to Turkic. In contrast, the Pathan levels are low, but biased toward Yakut ancestry. The ratio is very similar to the geographically close Hazara population, which is a clear Turco-Mongol ancestry in part, despite their adherence to a Persianate (Dari speaking) identity today. For me the point of curiosity is that the Pathan differ from the Baloch. The Baloch are in many ways just a more cosmopolitan spin on the Brahui, something that would make sense if the Iranian Baloch identity is an overlay upon a earlier Brahui layer. It would be interesting to know if the interaction with Turkic groups was historically known to be far less among the Baloch and Brahui than the Pathan, despite the geographically close position of these groups.
My analysis here is superficial. The analytic techniques aren’t too deep or informative. Rather, my intent was to push forward the project of exploring parahistorical dynamics through genetics. But “parahistory” I’m not talking counterfactuals, but rather what textually and even physically focused analysis of the past misses because of its methodological constraints. Would we know of the possibility of a relict Dravidian substratum in the hills of Balochistan deep into the medieval era if not for the fact that Brahui persists to this day? (it is giving way to Baloch even today) Many populations remain “dark” to textual records, or mentioned only as an aside to the “main stream” of military and fiscal concerns of rentier aristocracies or the poetic grandiloquence of literary elites. Archaeology in theory can compensate for this bias in the written word, but it is only an imperfect science in a positivistic sense. I am skeptical that archaeologists would have been bold enough to assert the existence of a Dravidian substratum in Balochistan even if there was a physical difference in the material objects. How would they even connect these people to Dravidian languages in the rest of South Asian anyhow?
As for the Turks and Pathans, some Turks did turn Pathan I believe. In fact one of the HGDP Pathans seems clearly to have been the product of recent admixture. That is not so surprising in light of history. Rather, it is critical to pin down specific values if we are ever to understand with any clarity the nature of ethnogenesis of the groups which met at the intersection of what the ancient Persians would have termed Iran, Turan, and Hind.
Comments are closed.