Earliest systematic coal exploitation for fuel extended to ~3600 B.P.

Abstract

Coal has long fueled human civilizations. The history of systematic coal fuel exploitation has been traced back to the late third millennium before present (post-2500 B.P.). Although sporadic combustion of coal for fuel was reported in some prehistoric archaeological sites, evidence for the systematic exploitation of coal for fuel before 2500 B.P. remains lacking. Here, we report comprehensive understanding for the earliest systematic exploitation of coal for fuel at the Jirentaigoukou site in Xinjiang, northwestern China, at ~3600 B.P. The main body of the site witnessed systematic exploitation of bituminous coals, illustrating a complete chaîne opératoire with selective mining, planned storage, and extensive combustion. Our results transform the knowledge of energy history by extending the upper limit of the systematic exploitation of coal for fuel by approximately a millennium, and provide a precedent of energy transition under intense conflict between social demand and environmental deterioration.

World’s earliest systematic exploitation of coal for fuel was extended by a millennium according to multidisciplinary evidence.

INTRODUCTION

One of the perennial quests in human history has been the search for new sources of fuel to exploit energy. Plants fueled the early anthropic controls of fire since ~790,000 B.P. (years before the present; taken as 1950 CE) (1, 2) or even earlier (3, 4). Since then, fuel has become an indispensable resource in subsistence, particularly considering that making fire was vital for early humans to better adapt to the living environment, and has probably profoundly affected their physiology and social organization (5–8). Subsequently, wood, charcoal, and dung were widely used as fuels. Coal became more important in the Chinese Song Dynasty (990 to 671 B.P.) (9) and played an obbligato role in the European Industrial Revolution (10).

Although coal has been used for making ornaments since the Late Paleolithic in Europe (11) and the Neolithic in northern China (12), the behavior of systematic exploitation of coal for fuel occurred much later. Evidence of sporadic lignite (weakly coalified coal) combustion for fuel has been found at several Late Paleolithic sites in Europe and North America (Fig. 1A) (13–15). However, none of these sites has yielded robust evidence or good chronologic constraints on how the coal was exploited by the local groups. On the basis of archaeological and textual evidence, the earliest reliable record for the systematic exploitation of coal for fuel is from the Chinese Han Dynasty (2152 to 1730 B.P.) (16), although people in the Warring States Period (2426 to 2171 B.P.) of China and Classical Greece (2460 to 2273 B.P.) seem to have mastered the knowledge of coal for fuel (17, 18). In Europe, people in Roman Britain started to burn coal for fuel in the Common Era (19). Some scholars asserted that coal was used in some locations during the Bronze Age of northern China (~4000 to 3000 B.P.) (20); nevertheless, a more detailed archaeological context and reliable chronology are required for further verification. Although recent research on biomarkers has revealed lignite combustion since ~3400 B.P. in the eastern Mediterranean (17), direct archaeological evidence for coal as fuel is absent. In summary, despite wide research interest, direct evidence of systematic coal fuel exploitation during prehistory was not found until the archaeological excavation of the Jirentaigoukou site (JRTGK) has been carried out (Fig. 1).

Fig. 1. Geographic setting of the archaeological sites and the coal outcrops mentioned in this study.

(A) Places known for the early utilization of coal fuel: (1) JRTGK (this study), (2) two Paleolithic sites in southern France (14, 15), (3) Lewisville Paleolithic site in North America (13), (4) sites with reported use of coal for fuel in central China (16, 20), (5) Roman Britain (19), and (6) two Late Bronze Age sites with evidence of lignite combustion in eastern Mediterranean (17). (B) Geographic setting of JRTGK and the relevant coal outcrops. (C to H) Images of the coal outcrops near the site, with dashed lines indicating their positions.

Abundant remains of coal have been unearthed at JRTGK (43°51′19″N, 82°47′07″E, 1226 m above sea level, also known as Jartai Pass or Goukou), a large Bronze Age settlement in the Upper Ili River Valley (UIRV; referring to the section located in China in this paper) of the central Tianshan Mountains (Fig. 1), providing a rare opportunity to explore when, why, and how local societies adopted coal as their major source of fuel. The major findings and preliminary understandings of the 5-year excavation program were introduced in several archaeological reports in Chinese (21–23), in which the discovery of coal exploitation for fuel has been mentioned and briefly discussed, but in-depth exploration remains lacking. On the basis of geochemical analyses of in situ coal remains and samples taken from adjacent coal outcrops, together with radiocarbon dates, soil geochemistry, and archaeobotanical evidence, this study reconstructed a comprehensive local history of the earliest systematic exploitation of coal for fuel. The combination of archaeological evidence and paleoenvironmental records provides a unique perspective on prehistoric human-environment interactions.

RESULTS

Radiocarbon dating and archaeological findings

Charcoals, human bones, and animal bones at JRTGK were sampled for radiocarbon dating for finer-grain chronology of the site, especially for fuel-related relics (data S1). Supplemented with published data (22, 24), the radiocarbon dates refined with Bayesian statistical modeling reveal a two-period human occupation of JRTGK at 4500 to 4300 B.P. and 3600 to 2900 B.P. (fig. S1). Only a small number of archaeological remains unearthed from JRTGK, including one burial and three charcoal kilns, were dated to the earlier period (~4500 to 4300 B.P.). These charcoal kilns are the earliest yet known and the sole residential remains of the Chalcolithic Age in Xinjiang (23). Numerous charcoals preserved in the layers of these kilns indicate that ligneous charcoal was the primary source of fuel (fig. S2). No evidence associated with coal exploitation during this period was found.

The main body of the site was dated to the later period (~3600 to 2900 B.P.), effected by the expansion of the Andronovo culture complex (21). A ~374 m² house foundation (F6) and a high-level square mausoleum with ~120 m side length (Gaotai) were found—probably the largest single building in prehistoric Xinjiang and the largest grave in prehistoric Eurasian Steppes thus far, indicating that JRTGK was a central settlement in UIRV (23). A large number of coal remains were unearthed from the cultural layers of this period, providing key evidence to study the technologies related to storage, processing, and usage (Fig. 2). Local excavation also revealed coal lumps piled in storage pits and houses, crushed by stone artifacts, and burnt as fuel in the hearths and smelting locations, as well as for making pottery molds. Well-preserved coal lumps were identified as bituminous coal, which is of higher rank than the lignite. These archaeological features reveal a complete chain of bituminous coal exploitation for various purposes, such as livelihoods and metallurgy. Charcoal remains were also unearthed from some of the archaeological units, such as house sites F1 and F7, but not as ubiquitous as coal.

Fig. 2. Typical archaeological remains related to coal exploitation at JRTGK.

(A) Coal ash unearthed from the floor of house site F1. (B) Coal pile unearthed from pit H112 in house site F25. (C) Coal lumps unearthed from house site F3. (D) Coal fragments incompletely burnt in stratum T2L3a. (E) Coal storage location near a furnace. (F) Stone artifact unearthed near a coal pile that was highly likely used for the coal fuel process.

Field survey and geochemical analyses of the coal

To better understand how prehistoric inhabitants exploited coal as their primary fuel resource, field survey was carried out to explore the coal outcrops nearby JRTGK. A total of six neighboring coal outcrops were spotted: The montane outcrops behind Qialegeer village, called Qialegeer 1, 2, and 3, are directly exposed on the hill slopes, while the riverine outcrops, namely, Jirentai, Yuxing, and Qishiertuan, are located on the fluvial plain of the Kashi River and covered by deposits 3 to 5 m thick (Fig. 1). These outcrops are the closest ones to the site, ~3 to 4 km as the crow flies. Considering that the long-distance transport of massive amounts of coal would have been inefficient and labor-consuming, these closest outcrops are the most reasonable coal sources for the site.

To further investigate the source of the coal, specimens sampled from these coal outcrops and from JRTGK were studied. Coal specimens taken from Piliqing, one of the nearest nonlocal modern coal mines, approximately 100 km from JRTGK, were also included in this study for comparative purposes. Geochemical analyses of Sr and Nd isotopes and the contents of carbon, nitrogen, and 38 trace elements [including rare earth elements (REEs)] were performed (Fig. 3, fig. S3, and data S2). Clustering analysis and principal components analysis (PCA) of trace elements were used to provide clearer insight into the relations among different coal specimens (Fig. 3, A and B). Although Sr and Nd isotopes are effective geochemical tracers, the information they provided here is limited, except for hinting the multiple sources of the coal used at JRTGK (Fig. 3C and data S2). Moreover, most of the indicators by themselves are insufficient to distinguish between coal specimens from the local outcrops and the Piliqing coal mine. By contrast, the clustering analysis did allow the specimens to be distinguished (Fig. 3A). Thus, from this perspective, most coal lumps unearthed at JRTGK show closer relations with coal specimens collected from the Yuxing, Qishiertuan, and Qialegeer 1 outcrops (Fig. 3A). In addition, the gangue specimens from Yuxing and Qishiertuan outcrops also show similarities with JRTGK (Fig. 3 and fig. S3), although the gangue specimens from other outcrops were not acquired. Coal specimens from the Jirentai outcrop vary distinctly in geochemical properties and show weak relation with other samples (Fig. 3 and fig. S3); thus, it appears not the coal source for JRTGK. The carbon contents of coal collected from the Qialegeer 2 and 3 outcrops are obviously lower (Fig. 3C), although these specimens look similar to other coal outcrop specimens that have higher carbon contents. In contrast, the coal lumps found at JRTGK have relatively consistent and good carbon contents (Fig. 3C). However, partially burnt coal specimens collected from JRTGK show relatively weak relations to the coal lumps from both the site and the coal outcrops (Fig. 3), which could have been influenced by the combustion process since the behavior of various elements during the combustion of coal is complex and diverse (25). Since all coal outcrops are of approximately equal distance to JRTGK and the terrain differences are unobvious (Fig. 1B), it seems that the people at JRTGK realized the unevenness in coal quality among different coal sources and consciously selected those with better quality (most likely from Yuxing, Qishiertuan, and Qialegeer 1 outcrops), which also indicates systematic exploitation.

Fig. 3. Geochemical analyses reveal multi-resource and selective mining at JRTGK.

In total, 47 coal/gangue specimens collected from JRTGK and 7 modern coal sources were measured. (A) Dendrogram generated by clustering analysis of 38 trace elements; samples were divided into six groups, which are delineated by the dashed boxes. (B) Scatterplot of PCA based on 38 trace elements. (C) Scatterplot of ⁸⁷Sr/⁸⁶Sr values versus carbon contents.

Detecting metallurgical activity at JRTGK

Unique metallurgical remains that represent a complete chain of bronze production, including bronzeware, ceramic molds, smelting furnaces, and slags, were unearthed from the later-period strata at JRTGK (fig. S2) (22, 23), indicating its key status in metallurgical activity during UIRV’s Late Bronze Age (3600 to 3000 B.P.). Supported by several local copper mines, the well-developed bronze industry at JRTGK is hitherto the earliest in Xinjiang (22). To further investigate the intensity of the metallurgical activity at JRTGK, we analyzed the elemental concentration of 288 soil samples from both natural soil and cultural layers of the two periods using x-ray fluorescence (XRF). The results show extensive enrichment of heavy metal elements in the soil around the smelting location. For instance, the enrichment of cuprum reaches a maximum of 50 times the background level near the smelting furnace at house site F7 (figs. S4 and S5 and data S3). However, samples from the earlier-period kilns show no obvious difference compared to the natural soil. This finding of metallurgical influence on the environment reflects the prominently increased human activity at the later period.

Wood fuel utilization at JRTGK

To achieve a more comprehensive understanding of the fuel utilization strategy at JRTGK, species identification of charcoal remains collected through handpicking from both earlier-period kilns and later-period house sites was performed (fig. S6E and data S4). All species identified were divided into three taxa: conifers, broadleaved trees, and shrubs (26). The earlier period saw wood fuel utilization with all taxa, namely, conifers (47.3%), shrubs (32.3%), and broadleaved trees (20.4%). Conifers are a good taxon for fuel due to their lower hardness and higher heat of combustion (27, 28); thus, it is reasonable to witness the preferential utilization of conifers in the earlier period at JRTGK. According to the charcoal remains from the later period, broadleaved trees became the dominant used taxon (94.2%), while percentages of conifers and shrubs prominently decreased (4.0% and 1.8%, respectively). However, the intense metallurgical activities during this period (21–23) should have required a vast consumption of fuel energy, making the abandonment of the superior conifers seemingly illogical. Moreover, the result of water flotation and sieving revealed that the density of the unearthed charcoal from this period (1.36 g/liter) was markedly lower than that of the earlier periods (25.98 g/liter) (29), implying a reduction in ligneous fuel exploitation. In short, the fuel exploitation strategy at JRTGK mainly transformed from conifer-preferred ligneous fuel in the earlier period to coal-dominated fuel in the later period. The mechanism lying behind such local transformation of energy resource is worth further discussion.

DISCUSSION

The earlier period of human activity at JRTGK occurred under more favorable climatic-environmental conditions, in which the climate was relatively warm and conifer forests in the Tianshan Mountains were widespread (fig. S6) (30, 31). Although archaeological evidence for this period in UIRV is sparse, the local Chalcolithic people undoubtedly adopted relatively primitive livelihoods in comparison to those in the later period, which may have resulted in insufficient social resilience when responding to climatic-environmental changes. The “4200 B.P. event,” known as a global climatic deterioration between the Middle and Late Holocene, has been noted as the driver for the recession of some ancient civilizations (32–34), and it is recorded in various sediments in Central Asia (fig. S6) (31). It is likely that the stress brought by the deteriorating climate and environment led to the gap in human activity during ~4200 to 3800 B.P. in UIRV (fig. S6).

Starting at ~3800 B.P., the Andronovo culture complex expanded rapidly on the Eurasian Steppes and introduced more sophisticated livelihoods into northwestern Xinjiang (35–38). Cultivation of multiple crops and domestication of multiple types of livestock, as well as complex bronze working, dramatically enhanced the complexity and resilience of local societies in UIRV (38), although the climate and environment were no longer as ideal as they were in the earlier period (fig. S6). On the one hand, the expansion of the local society and metallurgical activity (21–23, 38) required more supply of fuel resources. The relatively colder condition resulted in a receding and narrowing conifer forest belt in the Tianshan Mountains (30) and a reduction of the sum of trees and shrubs in the surrounding area (fig. S6), which reduced the supply of wood resources. Hence, we argue that the intense demand for fuel energy brought by a larger community and metallurgical production, together with a more limited supply of wood resources, escalated the conflict between local society and the environment, which finally triggered a break with local tradition in the form of adapting an alternative energy resource, the coal.

Since JRTGK is revealed as the sole site among the earlier and coeval sites in the neighboring regions to exploit coal (22), the local people may well be the pioneer to practice burning coal for fuel. The spontaneous ignition of exposed coal outcrops under hot weather is frequently witnessed nowadays; the same phenomenon may have inspired the local people that these black rocks could be burnt (23). The abundant coal resource around JRTGK (Fig. 1) would have resolved the energy crisis and promoted increases in the settlement size and metallurgical production. The bituminous coal used at JRTGK was of better quality than lignite, which could have been another powerful impetus in facilitating the development of the site into a regional center of UIRV by serving as a premium source of energy. The versatile usage in both daily life and metallurgical production at JRTGK (22) implies that coal, as a new type of fuel, was not necessarily controlled by any specific social class or craft production. Ultimately, the systematic exploitation of bituminous coal for fuel at JRTGK facilitated the formation of a unique complex local society (38).

Combining multidisciplinary research based on geochemical, environmental, and archaeological methods, our findings extend the human behavior of systematic exploitation of coal as fuel to ~3600 B.P., approximately a millennium earlier than previous estimation. Compared to the occasional burning of lignite in other earlier archaeological sites worldwide, the Late Bronze Age group at JRTGK exploited bituminous coal for fuel more actively and systematically, selecting the better-quality coal, planning specific coal storage, and using it for various purposes. This paper provides a case study in which prehistoric humans began to exploit a new energy resource under intense conflict between the booming human society and the deteriorating natural environment to achieve better sustainability. No archaeological evidence of coal fuel exploitation has yet been found within the neighboring regions post-2900 B.P., calling for more future studies to understand the temporospatial pattern of early use of coal for fuel.

MATERIALS AND METHODS

Archaeological context of the earliest coal evidence

JRTGK was discovered during the third national survey of cultural relics in 2008. Systematic excavations were performed in 2015, 2016, and 2018–2020 with the permission of the National Cultural Heritage Administration. The archaeological survey in JRTGK and neighboring regions revealed an area of ~80,000 m² of archaeological interests. The excavations targeted on the main body of JRTGK, a total area of nearly 6000 m², including 37 houses, over 300 other types of remains (e.g., hearths, ash pits, and smelting locations), and more than 1000 pieces of artifacts (including potteries, bronze wares, and stone artifacts) (22, 23).

Various types of coal remains, such as abundant coal lumps, partially burnt coal, cinder, and ash, and a small amount of coal gangue were excavated at JRTGK. They were widely distributed in almost all cultural layers of the later period at JRTGK, with some being concentrated in the houses or around the fireplaces. Coal remains were found across 111 excavation squares and 37 house sites of the later period, with 25 coal piles in relatively considerable size (e.g., length over 1 m). More intriguingly, coal lumps were often conveniently piled inside the house and nearby the fireplace or smelting location, whereas coal cinder and ash were distributed in the hearth or spread on the floor (Fig. 2) (21–23). Coal cinders were also found attached to the walls of smelting furnaces and ashes on the ground (fig. S2), indicating the direct use of coal for metallurgical activities.

Chronology

The chronology of JRTGK was constructed on the basis of absolute radiocarbon dating results and supplemented with stratigraphic correlations. Since previous radiocarbon dating work have established the absolute age of some relics in JRTGK (22, 24), our dating work focused more on fuel-related relics, while the ages of typical burials of different periods were acquired to further complete the chronology. To anchor the absolute chronology of the fuel-related relics, animal bones and charcoals from multiple archaeological units were sampled for radiocarbon dating. The results of radiocarbon dating are listed in data S1 and fig. S1. A total of 10 charcoal samples from different layers of the three earlier-period charcoal kilns (Y4 to Y6) were sampled for radiocarbon dating. Additionally, two charcoal samples for radiocarbon dating were acquired from later-period house sites F1 and F7, in which charcoal remains were relatively abundant and species identification was applied. For coal remains, direct radiocarbon dating is infeasible due to the serious old carbon effect. Moreover, other dateable materials, such as charcoals, charred seeds, or bones around the coal remains, were infrequent and may have been polluted by the coals. Thus, we used stratigraphic correlations to constrain the chronology of the coal remains. In total, 7 pieces of animal bones from different archaeological units were chosen for radiocarbon dating. According to stratigraphic relations, these archaeological units are homochronous with some of the units that contain coal remains (data S2) and cover the later-period stratum from the bottom to the top. One sample from ashpit H36 was dated to be much later, which may due to the strata disturbance and may not represent the age of that archaeological unit. Human bones were selected for radiocarbon dating to constrain the chronology of some typical burials. For M86, an earlier-period burial, samples were collected from human bones of two individuals (one adult and one child).

For pretreatment, bone collagen was extracted from all 14 bone samples according to the procedures in previous studies (39–41). These collagen samples and several charcoal samples were sent to Beta Analytic Inc. (Miami, FL) for accelerator mass spectrometry (AMS) ¹⁴C measurements following the ISO/IEC 17025:2017 standard. For the charcoal samples measured at the Key Laboratory of Western China’s Environmental Systems (Ministry of Education), Lanzhou University, a standard acid-alkali-acid pretreatment was performed (42). Then, the desiccated samples were compounded into graphite, and AMS ¹⁴C measurements were made on MICADAS 20. For calibration, the raw dating results were calibrated by an OxCal 4.4.4 online freeware with the up-to-date IntCal20 curve (43–45).

Bayesian chronological modeling

Bayesian chronological modeling was used to provide clearer boundaries of the initial and terminational ages of JRTGK instead of the conventional, morphology-based ages obtained from archaeology (43). The process was performed by the OxCal 4.4.4 online freeware (43, 44) and calibrated by the IntCal20 curve (45). The detailed operation of the OxCal online program for Bayesian modeling has been well described in a previous study (46). A total of 49 radiocarbon dates from JRTGK were collected from published results (n = 23) (22,24) and this study (n = 26) to achieve comprehensive coverage of the site. After eliminating the later Iron Age burials (n = 11), the accepted dates (n = 38) formed two phases: an early phase (n = 15) and a late phase (n = 23). The initiation age and the termination age of each phase are shown in fig. S1. The median of each age was adopted as the boundary of the two phases, based on current research (46, 47).

Coal quality identification

A group of 17 coal lump specimens collected from both JRTGK (n = 10) and coal outcrops (n = 7) were sent to the Xinjiang Institute of Coal and Gas Testing for quality identification based on the GB/T 5751–2009 Chinese national standard for coal classification, which is a standard modified from the international ISO 11760:2005 classification and is used nationwide in China. Both the ISO 11760:2005 and ISO 11760:2018 standards use the mean random reflectance ($\overline{R}$_r) to divide the quality of coals into three ranks: low rank (lignite and subbitumite), medium rank (bitumite), and high rank (anthracite). In comparison, Chinese national standard GB/T 5751–2009 also divides coals into these three levels but uses different indexes: the dry-ash-free volatile (V_daf) content and the transmittance (P_M). The GB/T 5751–2009 standard is an application-oriented classification that uses several indexes, including the caking index (G), Audibert-Arnu dilatation (b), and maximum thickness of the gelatin layer (Υ), to subdivide the rank of bituminous coal into 13 kinds. Except for unidentifiable samples that were eroded during the preservation process, the remaining coal samples are identified as 31BN (noncaking) or 41CY (long-flame) bituminous coal.

Geochemical analysis

Strontium (Sr) and neodymium (Nd) isotopes and trace elements (especially REEs) are effective geochemical tracers in geological research, while previous geochemical tracing of coal has focused on the processes of deposition and mineralization (48–52). Here, we attempt to provide a different approach for tracing the source of coal used at an archaeological site. Thus, multiple indexes, including ⁸⁷Sr/⁸⁶Sr, ¹⁴³Nd/¹⁴⁴Nd, concentrations of trace elements, and REE-specific indexes, such as the total concentration (Σ), ratio of the concentration between light and heavy rare earth elements (LREEs/HREEs), distribution pattern [North American shale composite (NASC)–normalized], cerium anomaly (δCe), and europium anomaly (δEu), were analyzed to comprehensively investigate the relations of coal between the archaeological site and different outcrops (data S2). In total, 47 coal specimens, consisting of coal lumps, partially burnt coal, and coal gangue, were collected from JRTGK and the coal outcrops mentioned above (Fig. 1). Among these, 20 specimens were sampled from different locations at JRTGK, while the remaining 27 specimens were collected from the coal outcrops, equally divided into nine groups. Plant roots and other impurities were eliminated from all specimens, and then the specimens were dried and ground into fine powders using a vibromill.

Sr and Nd isotope analyses were carried out at two institutions: (i) Key Laboratory of Surficial Geochemistry (Ministry of Education), Department of Earth and Planetary Science, Nanjing University, Nanjing, China (sample IDs: X1 to X20, Y1, Y4, Y7, Y10, Y13, Y16, Y19, Y22, and Y25) and (ii) Wuhan Sample Solution Analytical Technology Co. Ltd. (Wuhan, China) (the other 18 samples). Samples tested at both institutions were pretreated and measured using a Neptune Plus multicollector inductively coupled plasma mass spectrometry (MC-ICP-MS) instrument. The methods and standards involved have been reported in detail in previous studies (53–57). In total, 47 ⁸⁷Sr/⁸⁶Sr and 45 ¹⁴³Nd/¹⁴⁴Nd values were acquired, and two samples (Y11 and Y12) experienced failure during the Nd isotope analysis due to the low elemental concentration of Nd (data S2).

Trace element analysis of all 47 coal specimens was conducted on an Agilent 7700e laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) instrument at Wuhan Sample Solution Analytic Technology Co. Ltd., with an analytical uncertainty better than 5% (data S2). The methods adopted for the analysis followed Chinese national standard GB/T 14506.30–2010 and have been reported in detail in recent studies (58, 59). REEs were normalized to the NASC (60, 61) before plotting the REE distribution patterns (fig. S3).

Analyses of carbon and nitrogen contents, as well as stable carbon isotopes, were carried out at the same laboratory of Nanjing University mentioned in the above paragraph to further reveal the characteristics of different coal specimens. The analyses of carbon and nitrogen contents were conducted on an iCAP 6300 inductively coupled plasma optical emission spectrometry (ICP-OES) instrument. The δ¹³C_org values were acquired on a Picarro G2131-i δ¹³C analyzer with an ECS 4024 CHNSO elemental analyzer. Three in-house urea standards (δ¹³C_org = +11.71‰, −8.02‰, and −34.13‰) calibrated by the certified reference USGS40 were used for calibration, and the precision was better than 0.3‰. The pretreatment methods used before these analyses were as previously described (62, 63).

XRF measurement

Conducting XRF measurements on deposits in or around an archaeological site is an effective way to trace early metallurgical activity and has been successfully used in different areas (64–66). The 288 soil specimens analyzed in this study contain 20 specimens of natural sediment, 24 specimens from earlier-period kilns, 115 specimens from the smelting location (with one abnormal data point excluded), and 129 specimens from other locations from the later period (figs. S4 and S5). Natural sediments were sampled randomly from the topsoil around the site. Specimens of the earlier-period kilns (Y4 and Y5) were sampled from the bottom to the top. The specimens of the later-period cultural layers were collected from both the smelting location (around house site F7) and other locations (northwest section and southeast section of the site) at 1 or 2 m intervals precisely controlled by a real-time kinematic global positioning system (RTK-GPS). Pretreatment and measurements were performed in the Key Laboratory of Western China’s Environmental Systems (Ministry of Education), Lanzhou University, according to previously reported methods (66). An abnormal data point (XJ-17) among the 115 specimens of the smelting location was excluded due to its abnormal high values for all elements, which may be caused by the error state of the instruments. Stannum was not measured due to its excessively low concentration in soil, although it is also important in bronze production. Other heavy metal elements are sufficient to support the conclusion.

Constructing the SPD curve of UIRV

The summed probability distribution (SPD) constructed from radiocarbon dating data offers an effective way to reconstruct the intensity of human activity, especially the population in a certain area (67–69), which also promotes the study of human-environment interactions in the past (70, 71). By reviewing the published chronological data from archaeological sites in prehistoric UIRV, 110 radiocarbon dates were collected and input into the database to construct the SPD curve (22, 24, 72–79) (data S1). All radiocarbon dates were calibrated using the OxCal 4.4.4 program (42, 43) with the IntCal20 curve (45). The earliest radiocarbon result reported in UIRV came from a burial, namely, Dunna III M5, with four individuals being dated respectively (78, 80). A method refined by binning similar radiocarbon dates to avoid the SPD curve deviation due to archaeological bias (81) was also used. Here, raw radiocarbon ages from the same site and within 200 years were assessed as similar and were put into the same bin. In this method, the 110 radiocarbon dates were divided into 38 bins. The 38 binning results were input into the OxCal 4.4.4 program (43, 44) and calibrated with the IntCal20 curve to construct the SPD curve (fig. S6F).

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S6

Legends for data S1 to S4

References

Other Supplementary Material for this manuscript includes the following:

Data S1 to S4