Proto-Tocharians: From Afanasievo to the Tarim Basin through the Tian Shan


A reader commented recently that there is little information about Indo-Europeans from Central and East Asia in this blog. Regardless of the scarce archaeological data compared to European prehistory, I think it is premature to write anything detailed about population movements of Indo-Iranians in Asia, especially now that we are awaiting the updates of Narasimhan et al (2018).

Furthermore, there was little hope that Tocharians would be different than neighbouring Andronovo-like populations (see a recent post on my predicted varied admixture of Common Tocharians), so the history of both unrelated Late PIE languages would have had to be explained by the admixture of Afanasievo-related groups with peoples of Andronovo descent and their acculturation.

However, data reported recently by Ning, Wang et al. Current Biology (2019) confirmed that peoples of mainly Afanasievo ancestry – as opposed to those of Corded Ware-related ancestry expanding with the Srubna-Andronovo horizon – spread the Tocharian branch of Proto-Indo-European from the Altai into the Tian Shan area, surviving essentially unadmixed into the Early Iron Age.

This genetic continuity of Tocharians will no doubt help us disentangle a great part the ethnolinguistic history of speakers of the Tocharian branch of Proto-Indo-European, from Pre-Proto-Tocharians of Afanasievo to Common Tocharians of the Late Bronze Age/Iron Age eastern Tian Shan.

NOTE. Tocharian’s isolation from the rest of Late PIE dialects and its early and intense language contacts have always been the key to support an early migration and physical separation of the group, hence the traditional association with Afanasievo, a late Repin/early Yamna offshoot. Even with the current incomplete archaeological and genetic picture, there is no other option left for the expansion of Tocharian.

It is not possible to use the currently available ancestry data to map the evolution of Afanasievo ancestry, lacking a proper geographical and temporal transect of Central and East Asian groups. In spite of this, Ning, Wang, et al. (2019) is a huge leap forward, discarding some archaeological models, and leaving only a few potential routes by which Tocharians may have spread southward from the Altai.

NOTE. I have updated the maps of prehistoric cultures accordingly, with colours – as always – reflecting the language/ancestry evolution of the different groups, even though the archaeological data of some groups of Xinjiang remains scarce, so their ethnolinguistic attribution – and the colours picked for them – remain tentative.

A rough timeline of related archaeological sites from North Eurasia. Image modified from Yang (2019).


The recent book Ancient China and its Eurasian Neighbors. Artifacts, Identity and Death in the Frontier, 3000–700 BCE, by Linduff, Sun, Cao, and Liu, Cambridge University Press (2017) offers an interesting summary of the introduction of metalworking into western China.

Here are some relevant excerpts (emphasis mine):

Although [the Xinjiang] route is not uniformly agreed upon (Shelach-Lavi 2009: 134–46), this western transmission has been thought to have passed through eastern Kazakhstan, especially as it is manifest in Semireiche, with Yamnaya, Afanasievo (copper) and Andronovo (tin bronze) peoples (Mei 2000: Fig. 3). From Xinjiang this knowledge has been thought to have traveled through the Gansu Corridor via the Qijia peoples (Bagley 1999) and then into territories controlled by dynastic China. The dating of this process is still a problem, as the sites and their contents in Xinjiang are consistently later than those in Gansu, suggesting that the point of contact was in Gansu and that the knowledge then spread from there westward.

1. Eneolithic Altai

Afanasievo expansion ca. 3300-2600 BC. See full culture and ancient DNA maps.

The Afanasievo sites, as they are identified in Mongolia, for instance, make up an Eneolithic culture analogous to that of southern Siberia (3100/2500–2000 BCE) in the Upper Yenissei Valley that is characterized by copper tools and an economy reliant on horse, sheep and cattle breeding as well as hunting. (…) The Afanasievo is best known through study of its burials, which typically include groups of round barrows (kurgans), each up to 12 m in diameter with a stone kerb and covering a central pit grave containing multiple inhumations. In their Siberian context, burial pottery types and styles have suggested contacts with the slightly earlier Kelteminar culture of the Aral and Caspian Sea area.

The Afanasievo culture monuments, located in the northern Altai and in the Minusinsk Basin (the western Sayan), have been seen as analogous evidence for cross-Eurasian exchange. These complexes contain small collections of metal, and many of the items are made of brass, although golden, silver and iron ornaments were also identified. A mere one-fourth of these objects are tools and ornaments, while the rest consist of unshaped remains and semi-manufactured objects. Its metallurgical tradition has recently been dated by Chernykh to as early as 3100 to 2700 BCE (1992),making it more compatible chronologically with the early brass-using sites in Shaanxi mentioned above. Kovalev and Erdenebaatar have excavated barrows in Bayan-Ulgii, Mongolia, that have been carbon-dated to the first half of the third millennium BCE and associated by ceramic types and styles and burial patterns with the Afanasievo (Kovalev and Erdenebaatar 2009: 357–58). These mounded kurgans were covered with stone and housed rectangular, wooden-faced tombs that included Afanasievo-type bronze awls, plates and small “leaf-shaped” knife blades (Kovalev and Erdenebaatar 2009: Figs. 6 and 7).

They also excavated sites belonging to the more recently identified Chemurchek archaeological culture, located in the foothills of the Mongolian Altai (Kovalev 2014, 2015) (Fig. 2.6). These sites are carbon-dated to the same period as the Afanasievo burials or to c. 3100/2500–1800 BCE (six barrows in Khovd aimag and four in Bayan-Ulgo aimag). In the rectangular stone kerbed Chemurchek slab burials (Ulaaanhus sum, Bayan-ul’gi aimag and so forth), bronze items included awls; and at Khovd aimag, Bulgan sum, in addition to stone sculptures, three lead and one bronze ring were excavated (Kovalev and Erdenebaatar 2009: Figs. 2 and 3; Fig. 2.6). Although we will not know if they were produced locally until much further investigation is undertaken, these discoveries do document knowledge of various uses and types of metal objects in western and south central Mongolia. The types of metal items thus far recovered are simple tools (awls) and rings (ornamental?) not unlike those associated with Andronovo archaeological cultures as well.

This is a complex circumstance where archaeological evidence is not complete, but raises very important questions about transmission of metallurgical knowledge to and from areas in present-day China. In the 1970s some Afanasievo mounds were excavated in Central Mongolia by a Soviet–Mongolian expedition led by V. V. Volkov and E. A. Novgorodova (Novgorodova 1989: 81–85). Unfortunately, these mounds did not yield metal objects, only ceramics, but they show that the Afanasievo culture with the Eneolithic metallurgical tradition of manufacturing pure copper items had already moved east at least far as central Mongolia. In 2004, Kovalev and Erdenebaatar investigated a large Afanasievo mound, Kulala ula, in the extreme northwest of Mongolia, near the Russian border (Kovalev and Erdenebaatar 2009). There they found a copper knife and awl (Fig. 2.5). There are five C14 dates on wood, coal and human bones from this mound, which belong to the period 2890–2570 BCE. This shows that the Afanasievo culture were carriers of technology and produced artifacts in the first half of the third millennium BCE and that they also moved south along the foothills of the Mongolian Altai. Afanasievo culture in Altai and the Minusinsk basin is dated by C14 to 3600–2500 BCE (Svyatko et al. 2009; Polyakov 2010). In the north of Xinjiang in the Altai district, several typical egg-shaped vessels and two censers of Afanasievo types were found. Some of these have been obtained from the stone boxes (chambers of megalithic graves of the Chemurchek culture) (Kovalev 2011). Thus, the Afanasievo tradition of pure copper metallurgy must have spread to the northern foothills of the Tienshan Mountains no later than the mid-third millennium BCE. The links with Afanasievo and local cultures adjacent to and south of the mountains into present-day China can now be assumed.

Afanasievo – Chemurchek evolution ca. 2600-2200 BC. See full culture and ancient DNA maps.

2. Bronze Age Altai

Kovalev and Erdenebaatar (2014a) and later Tishkin, Grushin, Kovalev and Munkhbayar (2015) in Western Mongolia conducted large-scale excavations of megalithic barrows of the Chemurchek culture (dated about 2600–1800 BCE). This peculiar culture appeared in Dzungaria and the Mongolian Altai in the second quarter of the third millennium BCE and for some time existed together with the late Afanasievo culture, as evidenced by the findings of Afanasievo ceramics in Chemurchek graves, in the stone boxes. Unfortunately, in China we do not yet know of any metal object related,without doubt, to the Chemurchek culture. Kovalev, Erdenebaatar, Tishkin and Grushin found several leaden ear rings and one ring of tin bronze in three excavated Chemurchek stone boxes (Kovalev and Erdenebaatar 2014a; Tishkin et al. 2015). Such lead rings are typical for Elunino culture,which occupied the entire West Altai after 2400–2300 BCE (Tishkin et al. 2015). This culture had developed a tradition of bronze metallurgy with various dopants, primarily tin. Thus, the tradition of bronze metallurgy as early as this time could have penetrated the Mongolian Altai far to the south. In addition, in the Hadat ovoo Chemurchek stone box, Kovalev and Erdenebaatar discovered stone vessels refurbished with the help of copper “patches,” indicating the presence there of metallurgical production (Fig. 2.7) (Kovalev and Erdenebaatar 2014a). In one of the secondary

Chemurchek graves unearthed by Kovalev and Erdenebaatar in Bayan-Ulgi (2400–2220 BCE), a bronze awl was found (Kovalev and Erdenebaatar 2009). Kovalev and Erdenebaatar also discovered a new culture in the territory of Mongolia (Map 2.3), one that begins immediately after Chemurchek – Munkh-Khairkhan culture (Kovalev and Erdenebaatar 2009, 2014b). To date, about 17 mounds of this culture have been excavated in Khovd, Zavkhan, Khovsgol, Bulgan aimag of Mongolia. This culture dates from about 1800 to 1500 BCE, that is, contemporary with the Andronovo culture. Therefore, the Andronovo culture does not extend far into the territory of Mongolia. Three knives without dedicated handles or stems and five awls have been found in the Munkh-Khairkhan culture mounds (Fig. 2.8). All these products are made of tin bronze. (…) Additionally, eight Late Bronze Age burials (c. 1400–1100 BCE) were unearthed in the Bulgan sum of Khovd aimag and belong to another previously unknown culture called Baitag. And in the Gobi Altai, a new group of “Tevsh” sites dating to the Late Bronze Age were defined in Bayankhongor and South Gobi aimags (Miyamoto and Obata 2016: 42–50). From these Tevsh and Baitag sites, we see the expansion of burial goods to include beads of semiprecious stones (carnelian), bronze beads, buttons and rings and even the famous elaborate golden hair ornaments (Tevsh uul;Bogd sum;Uverkhanagia aimag) from the Baitag barrows (Kovalev and Erdenebaatar 2009: Fig. 5; Miyamoto and Obata 2016).

2.1. Chemurchek

About the Chemurchek culture, from A re-analysis of the Qiemu’erqieke (Shamirshak) cemeteries, Xinjiang, China, by Jia and Betts JIES (2010) 38(4):

The major characteristics of Qiemu’erqieke Phase I include:

  1. Burials with two orientations of approximately 20° or 345°.
  2. Rectangular enclosures built using large stone slabs. The size of the enclosure varies from a maximum of 28 x 30 m.*to a minimum of 10.5 x 4.4 m. (Figure 8, Table 2).
  3. *The stone enclosure located near Hayinar is the largest one at approximately 30 x 40 m. based on pacing of the site during a visit by the authors in 2008.

  4. Almost life-sized anthropomorphic stone stelae erected along one side of the stone enclosures (Lin Yun 2008).
  5. Single enclosures tend to contain one or more than one burial, all or some with stone cist coffins.
  6. The cist coffin is usually constructed using five large stone slabs, four for the sides and one on top, leaving bare earth at the base (Zhang Yuzhong 2007). Sometimes the insides of the slabs have simple painted designs (Zhang Yuzhong 2005).
  7. Primary and secondary burials occur in the same grave.
  8. Some decapitated bodies (up to 20) may be associated with the main burial in one cist.
  9. Bodies are commonly placed on the back or side with the legs drawn up.
  10. Grave goods include stone and bronze arrowheads, handmade gray or brown round-bottomed ovoid jars, and small numbers of flat-bottomed jars (Fig. 7).
  11. Clay lamps appear to occur together with roundbottomed jars.
  12. Complex incised decoration on ceramics is common but some vessels are undecorated.
  13. The stone vessels are distinctive for the high quality of manufacture.
  14. Stone moulds indicate relatively sophisticated metallurgical expertise.
  15. Artefacts made from pure copper occur.
  16. Sheep knucklebones (astragali) imply a tradition (as in historical and modern times) of keeping knucklebones for ritual or other purposes. They also indicate the herding of domestic sheep as part of the subsistence economy.
Chemurchek culture ca. 2200-1750 BC. See full culture and ancient DNA maps.

Chemurchek dating

Available evidence suggests that the date range for Qiemu’erqieke Phase I should fall from the later third into the early second millennium BC. There are several reasons to suggest that the time span is around the early second millennium BC. Lin Yun (2008) (…) maintains that the bronze artefacts found in Phase I show a greater sophistication in the level of copper alloy technology than that of the pure copper artefacts common to the Afanasievo tradition. On this basis it might be suggested that the Afanasievo could be considered to be Chalcolithic with a time span across much of the third millennium BC ( Gorsdorf et al. 2004: 86, Fig. 1). Qiemu’erqieke Phase I, however, should more properly be considered as Bronze Age.

Lin Yun also used the bronze arrowhead from burial Ml 7 to narrow down the date of Qiemu’erqieke Phase I. Two arrowheads were found in this burial, one of them leaf shaped with a single barb on the back (Fig. 7:4). A similar arrowhead, together with its casting mould, has been found at the Huoshaogou site of Siba tradition (Li Shuicheng 2005, Sun Shuyun and Han Rufen 1997), in Gansu province, northwest China, dated around 2000-1800 BC (Li Shuicheng and Shui Tao 2000) . This supports a date in the early second millennium BC for the Qiemu’erqieke arrowhead. The painted, round-bottomed jar from the Tianshanbeilu cemetery Qia Weiming, Betts and Wu Xinhua 2008: Fig. 7, bottom left) has been considered as a hybrid between the Upper Yellow River Bronze Age cultures of Siba in northwest China and the steppe tradition of Qiemu’erqieke in west Siberia (Li Shuicheng 1999). If this assumption is correct, the date of Tianshanbeilu, around 2000 BC, can be used as a reference for Qiemu’erqieke Phase I (Jia Weiming, Betts and Wu Xinhua 2008, Lin Yun 2008, Li Shuicheng 1999). Stone arrowheads found in Qiemu’erqieke Phase I also imply that the date is likely to fall within the earlier part of the Bronze Age as no such stone arrowheads have yet been found elsewhere in sites of the Bronze Age in Xinlang dated after the beginning of the second millennium BC.*
*For example Chawuhu and Xiaohe cemeteries (Xinjiang Institute of Archaeology 1999, 2003).

Pottery of Afanasevo and East European traits from the Chemurchek complex. Image modified from Kovalev (2017).

(…) Pottery “oil burners” (goblet-like ceramic vessels, possibly lamps) have been found in three traditions: Afanasievo (Gryaznov and Krizhevskaya 1986:21), Okunevo and Qiemu’erqieke. It is believed that this oil-burner found in Siberia and the Altai is a heritage from the Yamnaya and Catacomb
cultures (Sulimirski 1970: 225, 425; Shishlina 2008:46) in the Caspian steppe further to the west, but does not seem to exist in known Andronovo cultures.
The oil-burner tends to disappear after around 2300 BC during the mid-Okunevo period. It is, however, possible that the tradition continues longer in the Qiemu’erqieke sites.

The construction of the stone enclosures also reveals a close connection between Qiemu’erqieke Phase I and the mid and late Okunevo tradition (Sokolova 2007). Slab built stone enclosures emerged in both the Okunevo and Afanasievo traditions (Gryaznov and Krizhevskaya 1986:15-23, Kovalev 2008, Sokolova 2007, Anthony 2007:310, Koryakova and Epimakhov 2007). In the early Afanasievo the enclosure is circular with no cist coffin (Anthony 2007:310, Gryaznov and Krizhevskaya 1986:20), but in the early stage of the Okunevo square stone enclosures with a single cist burial are dominant. Square or rectangular stone enclosures are a marked feature of Qiemu’erqieke Phase I, suggesting temporal relationships between Qiemu’erqieke Phase I and the Okunevo. In Okunevo chronological group II, possibly with influence from the Anfanasievo, circular stone enclosures appeared in combination with rectangular enclosures within individual cemeteries, referred to by Sokolova (2007: table 2) as hybrid examples. By Okunevo chronological group III, rectangular stone slab enclosures with multi-burials emerged again. This is the dominant form in Qiemu’erqieke Phase I. Okunevo burial traditions changed again to single cist burials in the late stage around chronological group V ( Sokol ova 2007). A specific mortuary rite of decapitated burials exists in both the Qiemu’erqieke and Okunevo traditions (Sokolova 2007, Chen Kwang-tzuu and Hiebert 1995), as does the occasional occurrence of painted designs on the interior of the slabs forming the cists ( e.g., Khavrin 1997: 70, fig. 4; 77: tab. IV.5). Based on these comparisons, the date of Qiemu’erqieke Phase I may well parallel that of the Okunevo from at least chronological group II around 2400 BC (Gorsdorf et al. 2004: fig. 1).

Khuh Udzuuriin I-1 elite barrow (ca. 2470-2190 BC). Modified from Image modified from Kovalev (2014).

In addition to the pottery making tradition, the anthropomorphic stone stelae may also have earlier antecedents. In the Okunevo assemblage there are anthropomorphic stelae that are longer, thinner and more abstract than those of Qiemu’erqieke. There is no indication of such stelae in the Afanasievo tradition (Gryaznov and Krizhevskaya 1986:15-23). However, further to the west, anthropomorphic stone stelae are associated with the Kemi-Oba and Yamnya cultures around the third millennium BC (Telegin and Mallory 1994; Figure 13). Some major characteristics of these stelae such as the icons on the front face of the stelae (Telegin and Mallory 1994:8-9) also appear on stelae found in Qiemu’erqieke Phase I. Recalling the oil burners that may have been inherited from the Yamnya culture and which are found in the Afansievo, Okunevo and Qiemu’erqieke Phase I, it migh t be possible to speculate that Qiemu’erqieke Phase I has its origins even earlier than the first half of the third millennium BC. This idea has also been suggested by Kovalev ( 1999).

Despite the affinities with the Okunevo cultural tradition, Qiemu’erqieke Phase I appears to be a discrete regional variant. The ceramic assemblage shows traits unique to this cluster of sites, while the anthropomorphic stelae are also distinctive markers of this tradition.

Khuh Udzuur anthropomorphic stone stela, oriented toward the south – south-east. Image modified from Kovalev (2014).

3. Bronze Age Xinjiang

I recently reported on this blog the description of Xiaohe and Gumugou cemeteries from interesting Master’s thesis Shifting Memories: Burial Practices and Cultural Interaction in Bronze Age China: A study of the Xiaohe-Gumugou cemeteries in the Tarim Basin, by Yunyun Yang, Uppsala University, Department of Archaeology and Ancient History (2019).

It also offered a full summary of findings from prehistoric sites of Xinjiang related to the arrival of a cultural package from the Altai region, ultimately connected to Afanasievo. Relevant excerpts include the following (emphasis mine):

In Bronze Age Xinjiang, burials were diverse but also show some common features between different geographic sections. The main three mountains, including Kunlun Mountains, Tian Shan (mountains) and Altai Mountains, enclose the Tarim Basin, and the Dzungaria Basin, but leave the eastern part of the Tarim Basin and the south-eastern part of the Dzungaria Basin open (with easy access to the surroundings). The Hami Basin is located at the transitional area, connecting the two basins. Burials are mainly spread along the edge of the mountain ranges.

An assumption of the spreading/expansion routes stone burial construct.

3.1. The Lop Nur region

In the Lop Nur region, the Xiaohe cemetery (2000-1450 BCE) and the Gumugou cemetery (1900-1800 BCE) had many common features shared, and so is the Keliyahe northern cemetery:

  • Cemeteries were located in sandy areas;
  • Rectangular/boat-shaped wooden coffins with monuments of wooden planks or poles;
  • Coffins had no bottoms;
  • The dead were placed lying straight on the back;
  • The dead were commonly buried in single graves.

The Gumugou cemetery contained six special sun-radiating-spokes burial pattern in addition to the normal burials, which were similar to the wooden coffin graves of the Xiaohe cemetery.

NOTE. For more on Xiaohe and Gumugou, see the recent post on Proto-Tocharians. See other papers on the Andronovo horizon for other Early to Middle Bronze Age cultural groups less clearly associated with the Xiaohe horizon, like Hazandu, Xintala, or the Chust culture.

From Shuicheng (2006):

An assemblage of early bronzes had been recovered from northwestern Xinjiang and the periphery of Dzungaria 准噶尔 Basin. It comprises a variety of utilitarian tools and weapons, and a small number of apparels. These artifacts bear the stamps of Andronovo Culture in form, artifact type and decorative pattern. The metallographic analysis on selected artifacts indicates that they comprise mainly of tin-bronzes that contain 2–10% of tin. Moreover, the chemical compositions of these artifacts are similar to that of the Andronovo Culture. Latter date (first half of the 1st millennium BC) artifacts of the assemblage include a small number of arsenic bronzes. In all, during the period between the mid-2nd and mid-1st millennium BC, copper and bronze artifacts coexisted in this region, albeit tin-bronze comprised the majority. The composition of alloy did not show significant change over time. Some colleagues pointed out that the Nulasai 奴拉赛 site at Nileke 尼勒克 County in the Yili 伊犁 River basin of Xinjiang was the pioneer in the use of “sulphuric ore–ice copper–copper”technology. It is also the only early smelting site in Euro-Asia that arsenic ore was added to deliberately produce an alloy

Prehistoric cultures of Xinjiang during the Middle Bronze Age. See full culture and ancient DNA maps.

3.2. The Hami Basin-the Balikun Grassland

From Yang (2019):

The Hami Basin-the Balikun Grassland area is located at the eastern part of Tian Shan. The area is divided in a northern basin and a southern basin by the east-west stretch of the Tian Shan. In the Hami Basin-the Balikun Grassland area, the main type of burials were earth-pit graves in the early Bronze Age, and burials of stone-pit with barrows became more common in the late Bronze Age. The Hami-Tianshan-Beilu cemetery is a representative of the earth-pit graves. The features of the Hami-Tianshan-Beilu cemetery (2000-1500 bce) here were:

  • Rectangular earth pit graves;
  • The dead were often in a hocker position lying on one side;
  • Commonly a single dead in one grave.
The Balikun grassland today (source).

The Hami-Wubu cemetery (earlier than 1000 bce) and the Yanbulake cemetery (1200-600 bce) are representatives of another common earth-pit graves. Common features here were:

  • Rectangular earth pits, with two storeys and/or roofed with wooden boards;
  • The dead was placed in a hocker position lying on one side;
  • Mostly a single dead in one grave.

Later there appeared more stone-pit graves in this area, and the features can be summarized as:

  • Round burial mounds, commonly constructed by stones or a mix of stones and earth;
  • Burial mounds with a sunken top or a normal (dome) top;
  • The diameter of the burial mounds varied between 3 and 25.4 m (but not necessarily limited in this scope);
  • Circular or rectangular stone kerbs;
  • Rectangular stone pits, constructed by earth, or stones, or a mix of earth and stones;
  • Rectangular stone pits contained wooden coffins (represented by the Yiwu Baiqi’er cemetery).
Some representatives of stone burials in the Hami Basin – the Balikun Grassland in the Iron Age (Adapted from: Xinjiang 2011, 29-41). Image modified from Yang (2019).

In the Hami Basin, the Bronze Age cemeteries show common burial features like earth pits and hocker position of the dead. With similar pottery styles in the Hami-Tianshan-Beilu cemetery to those in the Machang and Siba cultures (Xinjiang 2011: 17), it suggests possible cultural influence or people’s migrating from the Hexi Corridor in the east.

In the Balikun Grassland, burials in an earlier time contained mostly earth-pit graves but also a small number of stone-pit graves. The pebbles were imbedded in the floors and the walls of the graves in a rectangular shape, e.g. the Balikun-Nanwan cemetery (1600-1000 bce). In a later time, there appeared huge burial mounds with a sunken top, and with the diameters of the burial mounds varying from 3 to 25.4 m, e.g. the Balikun-Dongheigou cemetery and the Balikun-Heigouliang cemetery. The Yiwu-Bai’erqi and the Yiwu-Kuola cemeteries contained either round stone burial mounds or circular stone kerbs on the ground surface. Considering the three burial elements including burial mounds, stone pits and circular kerbs, the later period cemeteries in the Balikun Grassland were actually similar to cemeteries from the southern edge of the Altai Mountain area.

From Shuicheng (2006):

The Nanwan 南湾 cemetery site at Kuisu 奎苏 Town, Balikun 巴里坤 (1600–1100 BC) also yielded an assemblage of early bronzes. The style of its early phase artifacts is similar to that of the burials distributed in the North Tianshan Route. Some sorts of cultural connection should have existed between the two.

The dates of Yanbulake 焉不拉克 Culture (1300–700 BC) are comparatively late. Its metallurgy was a continuation of the western China tradition. Artifact types include a variety of utilitarian tools, weapons and apparels.

Prehistoric cultures of Xinjiang during the Late Bronze Age. See full culture and ancient DNA maps.

3.3. The Turpan Basin-the middle part of Tian Shan

From Yang (2019):

Turpan Basin is located at the western part of the Hami Basin, and lies at the southern edge of the eastern Tian Shan. In the Turpan Basin-the middle part of Tian Shan area, the main representative of the Bronze Age cemeteries is the Yanghai Nr.1 cemetery. The features here were:

  • Elliptic earth pit graves, commonly covered by round logs on the top;
  • Some graves contained burial beds made of round logs or reeds;
  • The dead were mainly placed lying straight on the back;
  • Mostly a single dead in one grave.

In Iron Age, the stone burials became dominant, but the stone burials varied in different regions of the Turpan Basin-the middle part of Tian Shan area. Graves containing burial mounds, stone pit, and circular stone kerbs are represented by the Shanshan-Ertanggou cemetery, the Tuokexun-Alagou cemetery, the Urumqi-Chaiwobu cemetery and the Urumqi-Yizihu-Sayi cemetery, etc. The stone funeral construction features here are similar to those contemporary cemeteries in the Hami Basin-the Balikun Grassland area.

3.4. The southern edge of the western and middle part of Tian Shan

In the southern edge of the western and middle part of Tian Shan area, the main representatives of the late Bronze Age cemeteries are the Hejing-Chawuhu Nr.4 cemetery (around 1000-500 bce), the Hejing-Xiaoshankou cemetery, the Baicheng-cemetery, etc. The main burial features of the late Bronze Age and the early Iron Age cemeteries (see Fig.12) here were:

  • Burial mounds, constructed by stones or a mix of stones and earth;
  • Irregular circular or rectangular stone kerbs;
  • Stone pit graves in a bell-shape or a rectangular shape;
  • Stone pit graves constructed by imbedding pebbles or stone slabs in walls and floors;
  • The dead were often placed lying on their back with bent legs;
  • The dead were commonly reburied a second time with multiple burials.

From the late Bronze Age to the early Iron Age in this area, the burial traditions tended to be in a more varied way. In the stone burials with stone kerbs, there is a mixture of stone pit and earth pit graves. The burial features of the Iron Age cemeteries in this section were similar to those contemporary both in the Hami Basin-the Balikun Grassland area and in the Turpan Basin-the middle part of Tian Shan area.

From Shuicheng (2006):

The Chawuhu 察吾呼 Culture (1100–500 BC) distributes on the foothills between the middle section of the Tianshan Mountain Ranges and Tarim River. Its bronze assemblage comprises a variety of weapons, utilitarian tools and small apparels. They show no apparent temporal change in form and type through the four cultural phases. In addition, bronzes bear the Chawuhu characteristics were found in Hejing 和静, Baicheng 拜城 and Luntai 轮台 (Bügür). Yet, sites distributed along the Tarim River, such as Heshuo 和硕, Kuga 库车and Aksu 阿克苏, yielded remains of a bronze culture different from that of Chawuhu. Bronzes recovered include double-eared socketed axe, arrowheads, awls, knives, needles and bracelets. Their absolute dates have been estimated to be earlier than that of Chawuhu.

Prehistoric cultures of Xinjiang during the Early Iron Age. See full culture and ancient DNA maps

3.5. The Pamir Plateau

From Yang (2019):

A typical Bronze Age cemetery from the Pamir Plateau area is the Tashenku’ergan-Xiabandi cemetery (around 1000-500 bce). The burial features here were:

  • Mainly inhumations, but also a few cremations;
  • Burial mounds, constructed of stones;
  • Irregular circular or rectangular stone kerbs;
  • Mostly a single dead in one grave;
  • The dead was placed in a hocker position lying on one side.

The adoption of burial customs from the east supports the migration of Afanasievo-related peoples from the Tian Shan up to the Pamir Plateau, strongly influencing the findings of the Xiabandi cemetery, which has been dated from an early Bronze Age phase (ca. 1500-300 BC) to a late date up to ca. 600 BC.

While it is today unclear how far the Afanasievo admixture reached into the western Xinjiang, it seems that the Pamir Plateau remained culturally connected to neighbouring Andronovo-related cultures in pottery and metallurgical innovations, hence their language probably belonged – during most part of the Bronze and Iron Ages – to the Indo-Iranian branch, even though specific dialects might have changed with each new attested group.

In particular, it is possible that the early Andronovo groups related to the Xiaohe Horizon spoke Indo-Aryan or West Iranian dialects, while Saka-related groups replaced them – or an intermediate Tocharian-speaking group – with East Iranian dialects. A close interaction with West Iranian would justify the known ancient borrowings of Tocharian, although they could also be explained by contacts with Chust-related groups farther west. For more on this, see Ged Carling’s work on the different layers of Iranian loans.

Xinjiang BA/IA Summary

From Yang (2019):

In the early Bronze Age, there are distinct regional differences in the burial customs in and surrounding the Tarim Basin. At the southern edge of the Altai Mountains area, the burial customs included stone burial mounds, stone pit graves, circular or rectangular stone kerbs and stone human sculptures; the dead were placed lying straight on the back. In the Hami Basin-the Balikun Grassland area, the burial customs included earth pit graves; the dead were placed in a hocker position lying on one side. In the Turpan Basin-the middle part of Tian Shan area, the burial customs included earth pit graves; the dead were placed lying straight on the back. In the Lop Nur region, the burial customs included wooden coffins buried in sand; the dead were placed lying straight on the back.

But from the late Bronze Age to the early Iron Age, there was a common shift in burial customs from earth pit graves to stone burials in the Hami Basin-the Balikun Grassland area and in the Turpan Basin-the middle part of Tian Shan area. The main features of the stone burials include stone burial mounds, circular or rectangular stone kerbs, and the stone pit graves in the cemeteries. Similar stone burial customs commonly appeared at the southern edge of the western and middle part of Tian Shan area and the Pamir Plateau area in Iron Age. The burial features in most areas are in a mixture of both the earth pit graves and stone pit graves, especially in the Hami Basin-the Balikun Grassland area and the Turpan Basin-the middle part of Tian Shan area.


From Shuicheng (2006):

Historians of metallurgy conducted metallographic analyses on a sample of 234 metal specimens recovered from 16 localities in eastern Xinjiang. They concluded that the metallurgic industry in eastern Xinjiang could be roughly partitioned into three developmental phases. The early phase is represented by the burials distributed in the North Tianshan Route. The majority of the metal assemblage was tin-bronzes; however, copper and arsenic-bronzes maintained considerable proportions. The middle phase is represented by the burials at Yanbulake. During this phase, tin-bronze still maintained the majority; the proportion of arsenic-bronze increased, and some of them were high arsenic-bronzes. The late phase is represented by the burials at Heigouliang 黑沟梁. The composition of lead increased in the bronze alloy in the expense of arsenic. In addition, this phase witnessed the appearance of high tin-bronze that composed up to 16% of tin and the appearance of brass, that is, an alloy of copper and zinc. The bronze alloy consistently contained significant amount of impurities regardless of temporal difference. Casting and forging technologies coexisted throughout the three phases. The early bronzes (2000–500 BC) of eastern Xinjiang, in general, contained arsenic; however, the composition of arsenic was usually under 8%, but a few artifacts contained more than 20% arsenic. In all, arsenic had long been used in the alloy-forming of the early bronzes in eastern Xinjiang. Consequently, arsenic-bronzes were widely found in the prehistoric archaeology of the region. The artifact types, chemical compositions and manufacture techniques of the bronze assemblage of the burials of the North Tianshan Route are similar to those of Siba Culture, indicating that eastern Xinjiang had played a significant role in the East-West interactions.

An assemblage of early bronzes had been recovered from northwestern Xinjiang and the periphery of Dzungaria 准噶尔 Basin. It comprises a variety of utilitarian tools and weapons, and a small number of apparels. These artifacts bear the stamps of Andronovo Culture in form, artifact type and decorative pattern. The metallographic analysis on selected artifacts indicates that they comprise mainly of tin-bronzes that contain 2–10% of tin. Moreover, the chemical compositions of these artifacts are similar to that of the Andronovo Culture. Latter date (first half of the 1st millennium BC) artifacts of the assemblage include a small number of arsenic-bronzes. In all, during the period between the mid-2nd and mid-1st millennium BC, copper and bronze artifacts coexisted in this region, albeit tin-bronze comprised the majority.

Prehistoric cultures of Xinjiang during the Late Iron Age. See full culture and ancient DNA maps.

Tocharians in population genomics

Prehistoric population movements between the Altai and the Tian Shan are difficult to pinpoint, not the least because of the division of these territories among three different countries and their archaeological teams, only recently (more) open to the international scholarship.

The available schematic archaeological picture, where migrations could only be roughly inferred, has been recently updated to a great extent by Ning, Wang et al. (2019), whose genetic analysis of the samples is as thorough as anyone could have asked for, with a level of detail which matches the complex genetic picture of the region by the Iron Age.

As a summary, here is what they described about the samples from Shirenzigou (ca 400-200 BC), corresponding to the Iron Age populations of the Hami Basin-the Balikun Grassland area, and closely related to the preceding Yanbulake Culture:

As shown in Figure S3, the Steppe_MLBA populations including Srubnaya, Andronovo, and Sintashta were shifted toward farming populations compared with Yamnaya groups and the Shirenzigou samples. This observation is consistent with ADMIXTURE analysis that Steppe_MLBA populations have an Anatolian and European farmer-related component that Yamnaya groups and the Shirenzigou individuals do not seem to have. The analysis consistently suggested Yamnaya-related Steppe populations were the better source in modeling the West Eurasian ancestry in Shirenzigou.

Biplot of f3-outgroup tests illustrating the Kostenki14 and Anatolia_N like ancestries in Shirenzigou individuals. Most Shirenzigou individuals were on a cline with Yamnaya and European hunter-gatherer groups, lacking the European farmer ancestry as compared to the Steppe_MLBA populations such as Andronovo, Srubnaya and Sintashta [S1-S5]. Horizontal and vertical bars represent ± 3 standard errors, corresponding to form of outgroup f3 tests on the x axis and y axis respectively.

We continued to use qpAdm to estimate the admixture proportions in the Shirenzigou samples by using different pairs of source populations, such as Yamnaya_Samara, Afanasievo, Srubnaya, Andronovo, BMAC culture (Bustan_BA and Sappali_ Tepe_BA) and Tianshan_Hun as the West Eurasian source and Han, Ulchi, Hezhen, Shamanka_EN as the East Eurasian source. In all cases, Yamnaya, Afanasievo, or Tianshan_Hun always provide the best model fit for the Shirenzigou individuals, while Srubnaya, Andronovo, Bustan_BA and Sappali_Tepe_BA only work in some cases.

Table S2. P values in modelling a two-way (P=rank 1) admixture in Shirenzigou samples using each of the four populations (Bustan_BA, Sappali_Tepe_BA, Andronovo.SG, Srubnaya) together with Han Chinese as two sources [S6], Related to Figure 2. We used the following set of outgroups populations: Dinka, Ust_Ishim, Kostenki14, Onge, Papuan, Australian, Iran_N, EHG, LBK_EN.


In the PCA, ADMIXTURE, outgroup f3 statistics [see Figure S4], as well as f4 statistics (Table S3), we observed the Shirenzigou individuals were closer to the present day Tungusic and Mongolic-speaking populations in northern Asia than to the populations in central and southern China, suggesting the northern populations might contribute more to the Shirenzigou individuals. Based on this, we then modeled Shirenzigou as a three-way admixture of Yamnaya_Samara, Ulchi (or Hezhen) and Han to infer the source from the East Eurasia side that contributed to Shirenzigou. We found the Ulchi or Hezhen and Han-related ancestry had a complicated and unevenly distribution in the Shirenzigou samples. The most Shirenzigou individuals derived the majority of their East Eurasian ancestry from Ulchi or Hezhen-related populations, while the following two individuals M820 and M15-2 have more Han related than Ulchi/ Hezhen-related ancestry

It is unclear whether the Chemurchek population will show a sizeable local contribution from neighbouring groups. The fact that Okunevo shows 20% Yamnaya-related ancestry strongly supports the nature of neighbouring stone-grave-building peoples of the Altai and the northern Tian Shan as mostly Afanasievo-like, and the apparent lack of contributions of Srubna/Andronovo-like ancestry in the early Hami-Balikun stone burial builders also speaks for radical population replacement events reaching the areas south of Tian Shan, at least initially.

While ancestry cannot settle linguistic questions, it seems that nomads of the Gansu and Qinghai grasslands retained an ancestry close to Andronovo, whereas nomads of the Hami Basin-Balikun grasslands and related populations of Xinjiang remained closely related to Afanasievo. This doesn’t preclude that the ancestors of the Yuezhi became acculturated under the influence of peoples from eastern Xinjiang, but all data combined suggest an isolation of both populations – relative to other groups and to each other – and it is therefore more likely that they spoke Indo-Iranian-related languages rather than a language of the Tocharian branch.


In an interesting twist of events, despite the initially reported hg. R1b and Q, Tocharians from Shirenzigou actually show a haplogroup diversity comparable to that attested in other late Iron Age populations: a similar diversity is seen, for example, among Germanic, Baltic, and Balto-Finnic peoples of the Baltic region; among East Germanic or Scythians of the north Pontic region; or among Mediterranean peoples sampled to date. Iron Age peoples show thus a complex sociopolitical setting that overcame the previous patrilineal homogeneity of Bronze Age expansions.

PCA and ADMIXTURE for Shirenzigou Samples. Modified from the original to include in black squares samples related to Yamnaya. Modified from the paper to include labels of modern populations and a dotted lines with the cline formed by Shirenzigou, from (Yamnaya-like) Afanasievo to Central and East Asian-like populations. In red circles, samples with best fit for Andronovo-like ancestry. In green circles, samples with Han-related admixture.

M15-2 (with Han-related ancestry) is of the rare haplogroup Q1a-M120, while the samples with highest Steppe_MLBA-related ancestry are of hg. R1b-PH155, which points to their recent origin among Yuezhi, or to Hun-related populations showing an admixture related to the proto-historic nomads of the Gansu and Qinghai grasslands.

The expansion of Chemurchek-related peoples was probably associated more with hg. Q1a (dubious if it’s a Pre-ISOGG 2017 nomenclature, hence possibly Q1b), a haplogroup that might be found in Khvalynsk as a “significant minority” according to Anthony (2019), and it might also be attested in sampled individuals from Afanasievo in its late phase. This might be, therefore, a case similar to the early expansion of Indo-Europeans with R1b-V1636 lineages through the Volga – North Caucasus region, and of the later expansion with I2a-L699 lineages into the Balkans.

Haplogroup Q1a2-M25 is found in individual X3, whose Steppe ancestry is likely a combination of Afanasievo plus Andronovo-like ancestry heavily admixed with Hezhen/Ulchi-like populations, in line with the expected recent contacts with the neighbouring Xiongnu, Yuezhi, and other population movements affecting eastern Xinjiang.

Sample M4, which packs the most Afanasievo-like ancestry, is of hg. R1a-Z645, which – like sample M8R1 of hg. O – is most likely related to haplogroup resurgence events of local populations, which left the predominant Afanasievo-like admixture brought by builders of stone burials essentially intact, evidenced by the almost 100% of R1a found in the Xiaohe cemetery – and in most of the early Andronovo horizon – and among expanding Kangju and Wusun, as well as by the prevalence of hg. O among sampled East Asian populations.

A question that will only be answered with more samples is how and when the prevalent R1b-L23 and Q1b lineages among Afanasievo-related peoples began to be replaced to reach the high variability seen in Shirenzigou. Given the pastoralist nature of peoples around Tian Shan, the succeeding expansions of Proto-Tocharians, and the late isolation of different Common Tocharian groups, it is more than likely that this variability represents a late and local phenomenon within Xinjiang itself.

Peoples of Xinjiang during Antiquity. See full culture and ancient DNA maps.


Tocharians are one of the main pillars that confirm the Late Proto-Indo-European homeland of the R1b-rich populations of the Don-Volga region. There is already:

Just like the East Bell Beaker expansion from Yamnaya Hungary has confirmed that Corded Ware peoples did not partake in spreading Indo-European languages (spreading Uralic languages instead), data on the expansion of Tocharian speakers from Afanasievo to the Tian Shan was always there; population genomics is merely helping to connect the dots.

In summary, genetic research is supporting the expected linguistic expansions of the Neolithic and Bronze Age step by step, slowly but surely.


ASoSaH Reread (I): Y-DNA haplogroups among Indo-Europeans (apart from R1b-L23)


Given my reduced free time in these months, I have decided to keep updating the text on Indo-European and Uralic migrations and/or this blog, simultaneously or alternatively, to make the most out of the time I can dedicate to this. I will add the different ‘A Song of Sheep and Horses (ASoSaH) reread’ posts to the original post announcing the books. I would be especially interested in comments and corrections to the book chapters rather than the posts, but any comments are welcome (including in the forum, where comments are more likely to stick).

This is mainly a reread of iv.2. Indo-Anatolians and vi.1. Disintegrating Indo-Europeans.

Indo-Anatolians and Late Indo-Europeans

I have often written about R1b-L23 as the majority haplogroup among Late Proto-Indo-Europeans (see my predictions for 2018 and my summary of 2018), but always expected other haplogroups to pop up somewhere along the way, in Khvalynsk, in Repin, in Yamna, and in Bell Beakers (see e.g. the post on common fallacies of R1a/IE-fans).

Luckily enough – for those of us who want precise answers to our previous infinite models of Indo-European language expansions (viz. GAC-associated expansion, IE-speaking Old Europe, Anatolian homeland, Iran homeland, Maykop as Proto-Anatolian, Palaeolithic Continuity Theory, Celtic in the Atlantic façade, etc.) – the situation has been more clear-cut than expected: it turns out that, especially during population expansions, acute Y-chromosome bottlenecks were very common in the past, at least until the Iron Age.

Khvalynsk and Repin-Yamna expansions were no different, and that seems quite natural in hindsight, given the strong familial ties and aversion to foreigners proper of the Late Proto-Indo-European society and culture – probably not really that different from other contemporary societies, like the neighbouring Late Proto-Uralic or Trypillian ones.

Y-DNA samples from Khvalynsk and neighbouring cultures. See full version here.

Y-DNA haplogroups

During the expansion of early Khvalynsk, the most likely Indo-Anatolian culture, the society of the Don-Volga area was probably made up of different lineages including R1b-V1636, R1b-M269, R1a-YP1272, Q1a-M25, and I2a-L699 (and possibly some R1b-V88?), a variability possibly greater than that of the contemporary north Pontic area, probably a sign of this region being a sink of different east and west migrations from steppe and forest areas.

During its expansion, the Khvalynsk society saw its haplogroup variability reduced, as evidenced by the succeeding expansive Repin culture:

Afanasevo, representing Pre-Tocharian (the earliest Late PIE dialect to branch off), expanded with R1b-L23 – especially R1b-Z2103 – lineages, while early Yamna expanded with R1b-L23 and I2a-L699 lineages, which suggests that these are the main haplogroups that survived the Y-DNA bottleneck undergone during the Khvalynsk expansion, and especially later during the late Repin expansion. Nevertheless, other old haplogroups might still pop up during the Repin and early Yamna period, such as the R1b-V1636 sample from Yamna in the Northern Caucasus.

It is still unclear if R1b-L23 sister clade R1b-PF7562 (formed ca. 4400 BC, TMRCA ca. 3400 BC), prevalent among modern Albanians, expanded with Yamna migrants, or if it was part of an earlier expansion of R1b-M269 into the Balkans, and represent thus Indo-Anatolian speakers who later hitchhiked the expansion of the Late PIE language from the north or west Pontic area. The early TMRCA seems to suggest an association with Repin (and therefore Yamna), rather than later movements in the Balkans.

Y-DNA samples from Yamnaya and neighbouring cultures. See full version here.

‘Yamnaya’ or ‘steppe’ ancestry?

After the early years when population genetics relied mainly on modern Y-DNA haplogroups, geneticists and amateurs have been recently playing around with testing “ancestry percentages”, based on newly developed free statistical tools, which offer obviously just one among many types of data to achieve a proper interpretation of the past.

Today we have quite a lot Y-DNA haplogroups reported for ancient samples of more recent prehistoric periods, and they seem to offer (at least since the 2015 papers, but more evidently since the 2018 papers on Bell Beakers and Europeans, Corded Ware, or Fennoscandia among others) the most straightforward interpretation of all results published in population genomics research.

NOTE. The finding of a specific type of ancestry in one isolated 40,000-year-old sample from Tianyuan can offer very interesting information on potential population movements to the region. However, the identification of ethnolinguistic communities and their migrations among neighbouring groups in Neolithic or Bronze Age groups is evidently not that simple.

Yamnaya (Indo-European peoples) and their evolution in the steppes, together with North Pontic (eventually Uralic) peoples.Notice how little Indo-European ancestry changes from Khvalynsk (Indo-Anatolian) to Yamna Hungary (North-West Indo-Europeans) Image modified from Wang et al. (2018). See more on the evolution of “steppe ancestry”.

It is becoming more and more clear with each paper that the true “Yamnaya ancestry” – not the originally described one – was in fact associated with Indo-Europeans (see more on the very Yamnaya-like Yamna Hungary and early East Bell Beaker R1b samples, all of quite similar ancestry and PCA cluster before their further admixture with EEF- and CWC-like groups).

The so-called “steppe ancestry”, on the other hand, reflects the contribution of a Northern Caucasus-related ancestry to expanding Khvalynsk settlers, who spread through the steppes more than a thousand years before the expansion of Late Proto-Indo-Europeans with late Repin, and can thus be found among different groups related to the Pontic-Caspian steppes (see more on the emergence and evolution of “steppe ancestry”).

In fact, after the Yamna/Indo-European and Corded Ware/Uralic expansions, it is more likely to find “steppe ancestry” to the north and east in territories traditionally associated with Uralic languages, whereas to the south and west – i.e. in territories traditionally associated with Indo-European languages – it is more likely to find “EEF ancestry” with diminished “steppe ancestry”, among peoples patrilineally descended from Yamna settlers.

Y-DNA haplogroups, the only uniparental markers (see exceptions in mtDNA inheritance) – unlike ancestry percentages based on the comparison of a few samples and flawed study designs – do not admix, do not change, and therefore they do not lend themselves to infinite pet theories (see e.g. what David Reich has to say about R1b-P312 in Iberia directly derived from Yamna migrants in spite of their predominant EEF ancestry): their cultural continuity can only be challenged with carefully threaded linguistic, archaeological, and genetic data.