Journal of World Prehistory

, Volume 26, Issue 1, pp 1–24

The Neolithic Ceremonial Complex at Niuheliang and Wider Hongshan Landscapes in Northeastern China

Authors

    • School of Archaeology and Museology, Peking University
  • Andrew Bevan
    • Institute of Archaeology, University College London
  • Dashun Guo
    • Liaoning Provincial Institute of Cultural Relics and Archaeology
Article

DOI: 10.1007/s10963-013-9062-9

Cite this article as:
Zhang, H., Bevan, A. & Guo, D. J World Prehist (2013) 26: 1. doi:10.1007/s10963-013-9062-9

Abstract

This paper reviews the current evidence for settlement patterns and ceremonial activity amongst Hongshan Neolithic groups in northeastern China, with particular attention to the well-known ceremonial site at Niuheliang. We consider the location of Hongshan ceremonial sites in their wider landscape settings, arguing that such sites are a chronologically late stage of Hongshan ceremonial investment and that, within these broad complexes, the most impressive architecture and portable goods come from an especially late phase of activity. These more impressive localities were also particularly privileged places in terms of their access to major routes, specific kinds of local geology and integrated patterns of visibility. In contrast to more loosely organized Hongshan residential sites of similar size, ceremonial centres such as Niuheliang (and, within these, certain important sub-localities) clearly functioned as key mechanisms for social, political and regional stratification around roughly the mid 6th millennium BP.

Keywords

HongshanSacred landscapesSite location modelsViewshedsGIS

Introduction

Archaeological discoveries over the past two decades have made it very clear that a range of otherwise culturally distinct Neolithic societies across China undergo significant social and economic changes from about the mid 6th millennium BP. For example, apparently public structures and elaborate residences have been found in Late Yangshao period settlements in central China, while finely-decorated pottery forms and carved jade artefacts probably indicate elite burials at the Dawenkou, Lingjiatan and Songze cemeteries in eastern China. While it would be inappropriate to link the emergence of these regional Neolithic cultures with much later patterns of Chinese political unification, they are nonetheless the first observable developments in a trend towards considerably more complex societies in each of these sub-regions of China (Su 1997). In fact, comparably striking social changes can also been attested in northeast China from about the same period, and this paper focuses on the ritual and settlement landscapes of these Hongshan communities, which some have argued made important cultural contributions to the form and emphasis of later Chinese civilisation (Fig. 1). In particular, we consider patterns of settlement and ceremonial activity amongst these Hongshan groups, with special attention to the well-known ceremonial site at Niuheliang. Our main goal here is to consider the relationship between settled and ceremonial landscapes and, via a combination of updated archaeological evidence and spatial modelling, to more carefully unpick patterning at Niuheliang itself.
https://static-content.springer.com/image/art%3A10.1007%2Fs10963-013-9062-9/MediaObjects/10963_2013_9062_Fig1_HTML.gif
Fig. 1

Map of the main Neolithic cultural groups documented in China at c. 5500–5000 BP

Hongshan Settlement and Ceremonial Sites

Neolithic Hongshan culture was first characterised in the early 20th century and many archaeological excavations and surveys have since been devoted to its investigation. Archaeologically, the best-known and most diagnostic Hongshan artefacts are red painted or Z-striped ceramics, jade objects and certain multiform stone tools. Most Hongshan sites date to between c. 6500 and 5000 BP and are found across western Liaoning province and the southeastern part of Inner Mongolia. They thus cover a variety of landscapes, including the mountain ranges that stretch from northwest to southeast; the lower hill country to either side; various river valley systems; areas of arid desert to the north; the Bohai embayment in the south; and a large floodplain created by the lower Liao River in the east (Fig. 2; Guo 2005, pp. 1–34).
https://static-content.springer.com/image/art%3A10.1007%2Fs10963-013-9062-9/MediaObjects/10963_2013_9062_Fig2_HTML.gif
Fig. 2

Map of the main Hongshan sites

There are two well-known types of site associated with the Hongshan period: settlements and so-called ceremonial centres. Most of the settlements are located in river valleys created by the upper Liao River, the Daling River and their branches. Other settlement evidence has even been found in desert areas or in caves, but as yet only occasionally in the large eastern floodplain (Guo 2005, pp. 25–29), perhaps due to geomorphological factors. Some of our information about settlement sites comes from chance discoveries, and hence is only of limited analytical use with respect to overall regional patterning, but several systematic surveys and a few more detailed archaeological excavations provide deeper insights at several interpretative scales. Systematic regional surveys in the Hongshan area have used well-established, intensive or moderately intensive methods (for example, explicitly covered sub-areas, systematic line-walking, and/or formal surface collections), particularly within several river valleys, to explore community-level organisation and possible Hongshan demographic patterns (CICARP 2003; IMAT and MAB 2005; Peterson et al. 2010).

Despite this level of attention and fairly consistent archaeological retrieval, comparison between survey projects is complicated by further problems with regard to palimpsest remains covering both Hongshan and later periods, and the persistence of different methods for estimating variables such as site size. Even so, some general observations are possible. The vast majority of sites are small (expressing themselves as only a few hectares of surface scatter), and where excavated, produce no thick cultural deposits. This sits in contrast to Neolithic sites found in central China, along the Yellow and Yangtze rivers, at this time and suggests shorter-term occupations. In addition, there do seem to be a few rare examples of larger sites (e.g. >10 ha) that may indicate either a few bigger communities, perhaps of dozens of families each, or otherwise unusual patterns of occupation. Excavations of a few sites have revealed Hongshan houses that were often semi-subterranean, circular or square in shape, with timber frames and daub walls. The footprint of these houses is generally 20–30 m2, with only a few reaching up to 100 m2 (IMAT 1982; IMPICRA 1994, 2004). The few studies dedicated to plant and animal remains suggest a mixed household economy based on millet cultivation, wild food collection, pig husbandry and deer hunting, which would indicate exploitation of a variety of local landscapes (Guo and Ma 1985; Zhao 2005).

Although most people seem to have lived in small, potentially short-lived, villages, in partially-sunken dwellings, it can be argued that Hongshan settlements were also part of both local and regional aggregations each comprising many smaller sites. For example, at a coarse regional scale, more than 500 Hongshan sites have been identified within the c. 8,000 km2 of Aohan district alone, which greatly exceeds the densities recorded for other Neolithic societies in China by similar field methods. This dense aggregation is no doubt partially explained by the greater archaeological attention paid to this district, but it also probably suggests at least one, if not one of several, core regions of higher density activity (Shao 1995). At a finer local scale, spatial analysis of the regional survey data from the lower Bang River, Chifeng and Dongshanzui indicates that the Hongshan sites were loosely grouped into clusters spanning 2–8 km2 each (CICARP 2003; IMAT and MAB 2005; Peterson et al. 2010). However, it remains unclear whether we really should understand these aggregations as larger-scale political or economic entities, and if so, how such integration was accomplished, given the strong contrasts with other contemporary Chinese Neolithic cultures. The reinvestigation of ceremonial sites detailed below may provide further insight into these issues, by considering questions of location in the landscape, logistical organisation, and changing investment through time.

Hongshan ceremonial sites are also found in larger clusters of discrete localities. Ceremonial aggregations or centres of this kind include Niuheliang (牛河梁), Dongshanzui (东山嘴), Sijiazi (四家子) and Hutougou (胡头沟) amongst others (CBCC and LPICRA 2004; Guo and Zhang 1984; Shao 2004; Fang and Liu 1984), and are famous for their burials with jade artefacts and elaborate architectural additions. So far, more than 90 burials have been excavated at such centres, and at least 70 can be dated to approximately the same phase as Niuheliang (Hua and Yang 1998, p. 35). The burials typically involve grave pits cut vertically into the earth surface with slabs of limestone embedded on the edges, whose borders were often surrounded by painted pottery cylinders of uncertain function. The offerings in these burials were mainly made from jade river pebbles, suggesting much secondary access to this material rather than acquisition for primary outcrops. A further kind of monumental architecture involving circular or square altars of different sizes often appears together with these burial groups and may conceivably have been used for associated funerary ceremonies. In addition, at the site of Niuheliang, a semi-subterranean building, popularly known as the ‘Goddess Temple’, was unearthed, containing many ceramic nude female statues in different sizes (LPICRA 1986, and see below). The range of features at these ceremonial sites is one clear indication that the communities that produced them were able (a) to mobilise substantial labour to build these constructions, and (b) to develop the specialist crafting skills necessary to produce the jade carvings and painted, non-utilitarian ceramic forms. It is also obvious that the burials at these ceremonial sites cannot represent all segments of Hongshan society and that they were probably places for the interment of social, political and/or religious elites of some kind.

The known aggregations of ceremonial sites are mainly located on the southwestern edge of the known Hongshan settlement region. In most cases, they are found separated from the core zones of settlement, with Dongshanzui being the only probable exception to this rule. (At Dongshanzui, survey has certainly documented a combination of settlement and ceremonial remains in the same area [LPICRA et al. 2010; Peterson et al. 2010]. While there are as yet no published absolute dates or artefact assemblages to support either interpretation, it is worth considering the possibility that this site represents a palimpsest of different phases of activity within the longer Hongshan period, and that the settlement and ceremonial evidence may not be contemporary. In any case, this different pattern merits further investigation and Dongshanzui is discussed in a separate section below.)

Comparing the individual locations of the ceremonial sites to those of Hongshan settlements, the former are more likely to be on low hills in front of mountains which, as we argue more formally below, provided a specific environmental context whose meaning could be manipulated and enhanced in a variety of ways. In particular, the visual affordances (in the sense of Gibson 1986; see also Llobera 2001) offered at these ceremonial sites have for some time been emphasised by other commentators, with respect to the proposed symbolic meanings of certain mountains, and to patterns of intervisibility and alignment among sites (CBCC and LPICRA 2004). We reconsider some of these issues below, but it is worth simply noting at this point that such carefully selected, well-appointed ceremonial areas do appear to have been of great importance as sacred places with the potential to integrate more scattered Hongshan settled communities.

Niuheliang

Background

Niuheliang comprises a loose grouping of individual sites that are best thought of as a whole landscape of Hongshan ceremonial activity spread out over a series of separate low hills in western Liaoning province. The Niuheliang complex was first explored in the 1980s and consists of 16 main sites (LPICRA 1986, 1997a, b, c, 2001, 2008a, b; Li 1986; Wei 1994), with another 26 smaller sites identified in the same vicinity during a recent full-coverage survey project (Guo 2010a; Fig. 3). Together, these sites include: (a) over a dozen tomb groups; (b) several stone-set circular compounds and square altars; and (c) a large, irregularly shaped and partially sunken building known today as the ‘Goddess Temple’. The finds from these areas are exceptionally rich by Chinese Neolithic standards, with a wide range of jade objects, unusual decorated pottery forms and life-sized or larger clay statuary.
https://static-content.springer.com/image/art%3A10.1007%2Fs10963-013-9062-9/MediaObjects/10963_2013_9062_Fig3_HTML.gif
Fig. 3

Distribution map of the ceremonial sites at Niuheliang

We will now consider each of these three types of archaeological feature in turn. The tombs are rectangular and partially cut into the local bedrock. The biggest is as much as 4 m in length and width, and some of the larger tombs are cut almost 5 m deep into the hard bedrock. They often have a stepped interior, edged with large limestone blocks, and interior walls built with unbonded stacks of flat limestone pieces. Along the borders of many of the tombs were placed open-ended pottery cylinders with painted decoration. Offerings in the tombs were mainly jade objects in different shapes, such as bracelets, cuffs, ear ornaments, zoomorphic and human figurines (Fig. 4a, b).
https://static-content.springer.com/image/art%3A10.1007%2Fs10963-013-9062-9/MediaObjects/10963_2013_9062_Fig4_HTML.jpg
Fig. 4

Hongshan remains at Niuheliang: a a larger tomb; b a pottery cylinder

It has been suggested that the altars, most of which are stepped buildings with rammed earth interiors and stone-set exterior walls, and which terminated in either flat or rounded top platforms, were focal points for ceremonial activity. Some of them were built with red igneous rocks rather than the more popular white limestone of the surrounding buildings, and this would undoubtedly have made them visually distinct from their surroundings. A large pyramid-shaped building (Site 13) extends some 100 m in length at the base and is preserved up to 7 m in height. It was located separately from the tomb groups, but in terms of its design, it is similar to altars at tomb sites and might conceivably be regarded as another, bigger form of altar set apart from the rest (Fig. 5a, b).
https://static-content.springer.com/image/art%3A10.1007%2Fs10963-013-9062-9/MediaObjects/10963_2013_9062_Fig5_HTML.jpg
Fig. 5

Hongshan altars at Niuheliang: a Site 2; b Site 13

Finally, the so-called sunken building, or ‘Goddess Temple’, has clay-plastered walls, some of which are painted in red with geometric designs. The roof was probably arched, made with woven branches and clay, with dense embossed decoration. A great number of ceramic statues have been recovered from this building, including many broken nude female figurines of various sizes and some animal figures, such as dragons, boars and hawks (LPICRA 1986; Fig. 6a, b). A huge platform (40,000 m2) is located just behind the ‘Goddess Temple’, surrounded by a stone enclosure wall. At least three altars have been found on this platform, and some particularly large pottery vessels with bases up to 1 m in diameter, probably incense burners, were discovered at the same time (Hua 1994). The combination in one place of this platform with its group of altars and the ‘Goddess Temple’, undoubtedly indicates the primary importance of this location (Site 1) in the ceremonial activities of the entire Niuheliang site complex.
https://static-content.springer.com/image/art%3A10.1007%2Fs10963-013-9062-9/MediaObjects/10963_2013_9062_Fig6_HTML.jpg
Fig. 6

The ‘Goddess Temple’ (Site 1) at Niuheliang: a overview of the main structure; b clay female face found in the ‘Goddess Temple’

Chronology

It is now clear that not all the sites at Niuheliang were occupied simultaneously. Although the number of existing radiocarbon dates for Hongshan period sites is limited, recalibration of three from Niuheliang and one from Dongshanzui suggests (IACASS 1991, pp. 67, 76; Fig. 7) the possibility of a span of occupation over several hundred years, but places this at the end of the overall Hongshan period (clearly much later than the dates from settlement excavations at Xinglongwa; IACASS 1991, pp. 67, 76; Fig. 7). Typological studies of Hongshan pottery also suggest two or three sub-phases of Hongshan culture, with the Niuheliang sites probably dating to the last of these (Zhang 1991; Yang 1994; Zhang and Zhu 1994; Suo and Li 2007).
https://static-content.springer.com/image/art%3A10.1007%2Fs10963-013-9062-9/MediaObjects/10963_2013_9062_Fig7_HTML.gif
Fig. 7

Calibrated radiocarbon dates from Hongshan sites

A final important, but more tentative, chronological point is that excavations at certain sites within the Niuheliang complex seem to suggest a stratigraphic distinction between two discrete phases of tomb-building—known as the lower and upper burial deposits—although both belong to the Late Hongshan phase (LPICRA 1997b; Li 1986; Suo and Li 2007). The tombs buried in the lower deposits are comparatively small and are characterised by open-ended pottery cylinders placed in a ring around each tomb, with a great deal of broken rock filling inside. The central graves were rectangular with walls of flat limestone slabs, but without stone ledges or beds. These graves usually have a north–south orientation and most of the offerings are painted pottery cylinders (present at all such sites), or decorated urns with lids.

The built features of the tombs in the upper deposits are very different. Burial pits here are wholly covered with limestone slabs, including ledges and beds, and the orientation is primarily east–west. Mounds of stones, including some large blocks, are sometimes found above such graves, with painted pottery cylinders placed around the mound edges. The offerings within these seemingly later tombs are also entirely made of jade rather than pottery.

This apparent change in the character of tombs between lower and upper deposits ideally requires further confirmation via more explicit stratigraphic demonstration on a case-by-case basis, and further absolute dates, but seems likely on present evidence, and is important for our understanding of the development of the Niuheliang ceremonial complex through time. Painted pottery cylinders, which can easily be found in each individual locality at Niuheliang, also have observable differences in shape over this proposed timescale. Therefore we argue (and this is of relevance more generally at Hongshan ceremonial sites), that stylistic variation in cylinder sherds is a useful chronological indicator even for surface remains. On this basis, we can suggest that: (a) most of the altars were built during the period of the upper deposits and are only present at larger sites; (b) the construction of the ‘Goddess Temple’ may fall in between the periods of the lower and upper deposits; and (c) large tombs are only found in the period of the upper deposits and only at those sites with many smaller surrounding tombs.

Taken as a whole, this evidence indicates that sites in the Niuheliang area did not come into use at the same time. There were significant changes in the Niuheliang sacred landscape over time, and in the upper deposits period, class differentiation between sites becomes more obvious. The combination of tombs, altars and temples for ceremonial activities can only be found at larger sites. Meanwhile, there are also differences in the location of large as opposed to small sites at Niuheliang, which will be further discussed below.

Locational Patterns and the Material Landscape

The Niuheliang sites are found in a hilly landscape typical of the valleys of the upper Daling River, surrounded by mountains stretching from west Liaoning in the south to the western edge of the Hongshan zone. We start from the assumption that the locations for these sites are likely to have been selected with some reference to the demands of the ceremonial activities conducted there. Intuitively, and prior to the more formal analysis offered below, it seems that shallow ridge-tops of low hills close to the river valley were a preferred environment (the modern place name of Niuheliang itself means ‘Ox River ridge’ in Chinese), perhaps because these localities were well-drained and afforded good visual connections with others in the vicinity.

In addition, a great deal of labour, probably drawn from several Hongshan communities, was mobilised to build the tombs and associated ritual architecture, as well as to transport certain exotic building and burial materials from outside the region (Yan 2006). For example, some 30 m3 of gneiss bedrock was excavated with stone tools for one tomb alone at Site 16 (LPICRA 2008b, p. 9). Even so, it is worth pausing for a moment and putting this labour investment in wider comparative context. For example, Shelach et al. (2011) suggest that, in the same broad region, seemingly more elaborate stone fortifications could later be made by Lower Xiajiadian (4000–3200 BP) villages such as Sanzuodian (三座店), perhaps with a population of several hundred people, without the need for wider pooling of human labour. However, Lower Xiajiadian sites were less dispersed and probably longer-lasting than their Hongshan predecessors (e.g. Shao 1995), meaning that individual communities, or very limited cooperation amongst neighbouring ones, could bring together larger workforces. Moreover, with the possible exception of Dongshanzui (see discussion and note above), investment in Hongshan ceremonial sites occurred some distance away from settled areas and hence necessitated special purpose projects rather than everyday local additions. The suggestion that the burial customs at Niuheliang reflect influences from a range of neighbouring communities also argues in favour of this pooling of contributions (Yan 2006).

Turning to the effort and connections implied by portable material culture, most of the elaborate jade artefacts from the Niuheliang tombs were made from river pebbles of Xiuyan (岫岩) jade, which could only be found in the east of Liaoning and beyond the Hongshan region (Wang et al. 2007; Wen 1990). Likewise, although plenty of painted pottery cylinders have been found in the Niuheliang area, no ceramic workshops or kilns have been identified and it seems likely that these important objects were also manufactured elsewhere and then transported to Niuheliang. The most important access routes to the Niuheliang area are a series of inter-linked river valleys. The Daling River flows from west to east across the mountains, with many branch streams following a northeast to southwest direction that made them convenient for both land and river transportation. In fact, even today, the route of the railway from Beijing to Shenyang and a major national road (G101) follows one of these branches of the Daling River (known as the Ox River). Local geological variability seems to have been important, with a particular kind of white limestone being the typical building material for coffins and altar bases; several plausible local source outcrops exist in the Niuheliang area (with a long, later history of quarrying). The large limestone slabs used in the Niuheliang monuments are typically about 40 cm in length and of a size that one person could carry (LPICRA 1986, p. 7). Gneiss, however, is the most common bedrock in the immediate vicinity of Niuheliang (Mo et al. 2002, p. 179)—as we shall see below, most of the Niuheliang sites are found on this geology, being either set into the overlying soil or, in the case of the larger tombs, actually cut down into the bedrock itself.

The Visual Landscape

The visual affordances offered by certain landscapes have long been emphasised in archaeological studies of sacred places, particularly with regard to the way such visual properties might be further invested with social and cultural meanings (e.g. Fraser 1988; Tilley 1994; Zubrow 1994). Two inter-related features of the visual landscape at Niuheliang that we consider further below are (a) views of natural scenes, and (b) views among the artificial monuments. To the modern viewer, the most prominent general feature in the Niuheliang area is the sharp line of mountains rising to the southeast. Much later on in Chinese history, high mountains were traditionally and consistently regarded as sacred places where heaven and earth met and where various forms of shamanic and/or other ritual practice might occur (Barnes 1999, pp. 117–119). For example, later Chinese emperors typically held ceremonies on the top of famous mountains to legitimate their rule, particularly at the moment of succession. This connection between heaven-and-earth cosmologies, animal forms and the exercise of religious power (e.g. captured by the later notion of Wuism) is something that several authors have made very strongly with respect to Niuheliang (Rong 1993; Gu 2006; Cao 2007; Fu 1990, 2000), on the basis of finds such as jade artefacts thought to be in the form of specific animals, of a magical or religious human figure and of clouds (Fig. 8a, b), as well as the undeniable importance ascribed to depositing circular ceremonial ceramics on the tops of hills. In our view, while shamanic practice may well have been important as a source of local secular and ritual authority, and the possibility of conceptual continuity with much later Chinese ritual or cosmology is certainly worth floating as an idea, both remain extremely difficult to confirm on the basis of the existing evidence. Regardless, the central role of nature worship and of investment in certain symbolically-charged images are the two factors of Niuheliang portable and fixed material culture that it is impossible to ignore, and this prompts us to explore visual affordances further in the modelling that follows below.
https://static-content.springer.com/image/art%3A10.1007%2Fs10963-013-9062-9/MediaObjects/10963_2013_9062_Fig8_HTML.jpg
Fig. 8

Jade objects from Niuheliang tombs: a human figure, sometimes thought to be a shaman; b possible cloud shape

Likewise, since the archaeological discovery of the Niuheliang complex, far more specific visual meanings have been regularly ascribed to its mountain skyline. In particular, great importance has hitherto been placed on a more isolated and distinctively-shaped mountain to the southwest of Niuheliang which is part of what is, geologically-speaking and in local terms, a unique unit of porphyritic bedrock. The local name for this feature is Mulan Shan (木栏山), ‘Wood Column Mountain’, but regional archaeologists have suggested alternative names such as ‘Boar’, ‘Pig-Head’ or ‘Bear’ Mountain (Barnes and Guo 1996), based on (a) the resemblance of the mountain to the top of an animal head (as seen from the front), (b) an awareness of the long tradition of worshipping one or more of these animals in this region, and (c) the prevalence of animal-shaped Honghan artefacts at Niuheliang (especially a famous carved jade loop in animal form that was found here: Fig. 9a, b; CBCC and LPICRA: figure 95). These modern archaeological names are therefore meant to suggest the kinds of meanings that Hongshan communities may also have ascribed to the feature, and to propose some clear interpretative links with later Chinese or northwest Asian cosmologies. Again, it remains difficult to do more than speculate about these tight symbolic associations, but it seems likely that the combination of this particular mountain’s unusual shape, unusual bedrock and prominent skyline vista may indeed have encouraged some quite specific concepts of place.
https://static-content.springer.com/image/art%3A10.1007%2Fs10963-013-9062-9/MediaObjects/10963_2013_9062_Fig9_HTML.jpg
Fig. 9

Animal imagery associated with Niuheliang: a Mulan Shan, a mountain whose outline on the horizon has been likened by some to a boar, pig or bear; b jade object from a Niuheliang tomb, thought by some to be in the shape of boar or bear

In any event, the built monuments at Niuheliang may also have worked amongst themselves to construct a local cultural landscape of particular places and special meanings. For example, the Niuheliang structures emphasise visual contrast and colour: the larger tombs were mainly made from white limestone, with red stone sacrificial altars nearby and thick red cylindrical pots placed above them in rows that stood out from the background soils and bedrock (Guo 2010b). Most of the large sites within the overall complex are also fairly evenly distributed along the so-called Niuheliang ridge, which stretches from northeast to southwest. The most prominent place on this ridge is at the northern end where the ‘Goddess Temple’ is located, and the huge platform behind the temple undoubtedly re-emphasizes the importance of this northern section for the entire surrounding landscape. We further consider patterns of inter-visibility amongst these sites via formal modelling below.

Multivariate Spatial Analysis

A Locational Model

We can test some of the hypothetical locational properties mentioned above in a more formal way via multivariate spatial analysis. Two separate groups of data will be modelled—(a) all ceremonial sites in Niuheliang, and (b) a subset of larger sites from the proposed ‘upper deposits’ phase—in order to consider any possible differences between them. These groups will be compared via multivariate logistic regression, in which the presence data are sites and the absence data are a set of random points placed within the area of systematic, and in some sense ‘full coverage’, exploration at Niuheliang.

The variables chosen as independent covariates below are clearly not an exhaustive set, but are intentionally those that test informal associations either made by the authors in the field or already published in the interpretative literature. Table 1 summarises these variables, the data sources they are constructed from, and any statistical pre-treatment they have received. Each variable was first explored for its distributional properties, its interactions with other possible covariates, and for its significance as a univariate predictor. Thereafter, promising covariates were transformed if necessary and included in the multivariate model (with a final model being selected via stepwise minimisation of an Akaike Information Criterion, AIC). The 20 m digital elevation model (DEM) used here was interpolated from contours at 10 m intervals: all covariates were rasterised to the same resolution. One variable worth slightly more extended consideration is that of visibility. There are a range of modern methods for calculating and comparing the viewable area associated with a particular point in the landscape, known as its viewshed. The viewshed of the mountains that we use here is simply a binary map recording the areas from which at least one of the main mountain peaks to the southeast of Niuheliang could be seen in clear weather conditions. Likewise, a similar kind of viewshed was calculated and used further below to consider which sites are intervisible with the ‘Goddess Temple’ or the pyramid-shaped building.
Table 1

Summary regression results

 

(Constant)

Low elevation

Slope

Southeast aspect

Distance to river crossing

Ridge landform

Proximity to gneiss bedrock

Mountain view

Univariate p value

0.026

0.001

0.048

0.000

0.004

0.000

0.004

Model A

        

 p value

0.003

0.003

0.004

0.018

0.002

0.014

 Coefficient

−3.998

2.648

−0.001

2.101

2.434

2.519

Model B

        

 p value

0.416

0.039

0.026

0.016

0.005

0.011

 Coefficient

−0.774

−0.122

1.682

−0.001

2.451

1.876

Significant variables are shown in the first row of Table 1, including low elevation, slope, more southeasterly aspects, distance to the main river crossing, a fuzzy classification of ridge-like landforms, distance into/out of units of gneiss bedrock, and intervisibility with the mountainous area to the south. In particular, this last variable is treated carefully below: the exploration of sacred landscapes via GIS-based viewshed analysis has attracted many archaeologists in recent years, as this technique has, rightly or wrongly, been touted as a GIS-led method that might go beyond the consideration of straightforward environmental factors and look at issues of human perception (see Zubrow 1994; Lake and Woodman 2003; Llobera 2003). In order to test whether the view of mountains is truly significant for the choice of site location, two separate regression models have been compared: Model A begins with all significant univariate variables; Model B excludes the mountain view variable.

Table 2 indicates the signed correlations between each variable chosen for Model A. One point to note is that the mountain views variable is partly correlated with the aspect to the southeast (r = 0.551 or r2 = 0.304), because the entire valley opens out to the southeast. Another variable, distance to the main river crossing, is also slightly correlated with the mountain views (r = −0.315 or r2 = 0.099). These correlations urge a little caution in interpreting the results as evidence for explicit causal mechanisms, and we return to this issue below, but even so, the overall positive percentage of correct predictions (Model A 88.4 %, Model B 74.4 %), Akaike Information Criteria (Model A 84.94, Model B 87.94) and Receiver Operating Characteristic (ROC) curves (Model A 0.884, Model B 0.871) all suggest that Model A is slightly more effective than Model B, and hence that mountain views, whilst undeniably important in tandem with southeasterly aspects, may also be independently influential.
Table 2

Correlation matrix for model covariates

 

Southeast aspect

Distance to river crossing

Ridge landform

Proximity to gneiss bedrock

Mountain view

Southeast aspect

1.000

−0.315

0.054

0.366

0.551

Distance to river crossing

−0.358

1.000

−0.056

−0.108

−0.315

Ridge landform

0.054

−0.056

1.000

0.350

0.044

Proximity to gneiss bedrock

0.366

−0.108

0.350

1.000

0.182

Mountain view

0.551

−0.315

0.044

0.182

1.000

Models A and B addressed all sites in the Niuheliang complex. As a second stage of analysis, we also sought to consider only the larger built monuments, using the same approach. The reduced number of sites and non-sites in the sample still leads to predictor variables that are very significant, but with some continuing concerns over interdependence. Below we isolate patterns of intervisibility in other ways to confirm that it was indeed important for these larger sites, but we can also note in passing that compared to the smaller sites, the later-built larger ones were located in particularly homogenous places with a range of intersecting landscape affordances.

Intervisibility Among Sites

So far in terms of visibility, we have considered the rather simple case of the intervisibility of Niuheliang sites with the main mountainous area to the south. Given the above suggestion that this is a useful predictive factor, and more tentatively, that it might be an intentional feature of site location, we can also consider other aspects of the visual landscape. First of all, just how unusual is the overall amount of the landscape that you can see from the Niuheliang sites? Using a coarser 50 m DEM (interpolated from contours at 20 m intervals) with an appropriate buffer zone to control for edge effects (see Lake et al. 1998), viewer and target heights set to 1.5 m, and a background 10 % sample of random sites, we can calculate viewsheds that confirm, as the above regression models suggest, that these southeast-facing sites do indeed have better overall landscape visibility (Fig. 10). Furthermore, the vistas available from the larger sites are also larger than those from the smaller ones (p < 0.05; Fig. 11). We therefore suggest that it was partly the increased visual opportunities at certain already-active ceremonial locations at Niuheliang that prompted their more substantial investment with altars and elaborate burials.
https://static-content.springer.com/image/art%3A10.1007%2Fs10963-013-9062-9/MediaObjects/10963_2013_9062_Fig10_HTML.gif
Fig. 10

Cumulative viewshed analysis (CVA) of Niuheliang sites: a CVA from larger sites at Niuheliang; b CVA based on 10 % random sample cells

https://static-content.springer.com/image/art%3A10.1007%2Fs10963-013-9062-9/MediaObjects/10963_2013_9062_Fig11_HTML.gif
Fig. 11

Cumulative percentage of visible sites in Niuheliang with 10 % background sample

Taking this a step further, we can tentatively explore the visual network between the larger sites in a bit more detail. For example, intervisibility among these larger sites is also seemingly of some importance: Site 1 is where the ‘Goddess Temple’ is located, associated with large altars surrounded by a stone enclosure wall, while the so-called pyramid building, probably a huge altar, is Site 13. These two are distinctive not only for their exceptional finds but also for their special locations on hilltops where they are very often within visual range of the other sites. More formally, we have been able to construct a visual network, this time using the finer-scale 20 m DEM and recording intervisibility between the 16 large sites (with viewpoint height set to 1.5 m and target to ground level). To explore the degree to which the results might be sensitive to inaccuracies in the elevation model or intervening vegetation, we also re-ran the analysis 100 times, in each case randomly perturbing all cells in the DEM by ±1 m before re-running the viewsheds. The resulting visibility network shown in Table 3 is therefore probabilistic, with a 1 indicating that that the target site can always be seen from the viewpoint site in all runs, a 0 indicating that it never can, and a range of possible values in between indicating uncertainty. Thereafter, we used a Monte Carlo simulation—random re-labelling of two sites representing bigger monuments, such as the pyramid building and ‘Goddess Temple’, while keeping the visual relationships fixed. The results suggest at least the possibility that a visual focus on Sites 1 and 13 was important in determining which sites became larger in the later phases at Niuheliang (p = 0.11, using the percentile method and 10,000 iterations, and summing the probabilities of seeing Sites 1 and 13 from each of the other sites). However, there are clearly considerable uncertainties associated with only minor changes in intervening landscape heights (and by extension, probably also viewpoint and target heights), and further analysis with a much higher resolution elevation model and a palaeo-environmental reconstruction would be desirable.
Table 3

Probabilistic visibility network for large sites at Niuheliang

 

From 1

From 2

From 3

From 4

From 5

From 6

From 7

From 8

From 9

From 10

From 11

From 12

From 13

From 14

From 15

From 16

1

NA

0

0.57

0

0.72

1

0.04

0

0

0.73

0.07

0.86

0.83

0.9

0

0.93

2

0

NA

0.19

0

0

0

0

0

0.4

0

0

0

0

0

0

0

3

0.92

0.04

NA

0

0

0

0

0.86

0.83

0.03

0

0.07

0

0.08

0

0

4

0

0

0

NA

0.56

0.32

0.38

0

0

0

0

0

0.16

0.8

0

0

5

0.74

0

0.04

0.25

NA

0.17

0.47

0

0

0.31

0.55

0.6

0.2

0.74

0

0

6

0.82

0

0

0.01

0

NA

0.38

0

0

0

0

0

0

0.01

0

0.05

7

0

0

0

0.07

0.07

0.03

NA

0

0

0.13

0.36

0.37

0.36

0.35

0

0.53

8

0

0

0.66

0

0

0

0

NA

0.18

0

0

0

0

0

0

0

9

0.02

0.23

0.55

0

0

0

0

0.67

NA

0

0

0

0

0

0

0

10

1

0

0.38

0

0.88

0.07

0.84

0

0

NA

0

0

0

0.64

0

0

11

0

0

0

0

0.8

0.35

0.96

0

0

0

NA

0.4

0

0.73

0

0.47

12

1

0

0.48

0

0.97

0.41

0.97

0

0

0

0.62

NA

0

1

0.11

0.43

13

0.76

0

0

0.71

0.7

0.22

0.7

0

0

0.04

0

0.09

NA

0.99

0.9

0.99

14

1

0

0.7

1

1

0.57

0.94

0

0

0.98

1

1

1

NA

0.55

1

15

0

0

0

0

0

0

0

0

0

0

0

0

0.37

0.55

NA

0.55

16

1

0

0

0

0.02

0.64

0.97

0

0

0

0.97

0.94

0.99

1

0.94

NA

Other Hongshan Ceremonial Sites

Sijiazi

The Hongshan ceremonial sites at Sijiazi are located 60 km to the northeast of Niuheliang in southeastern Inner Mongolia. The character of the Sijiazi landscape is similar to that at Niuheliang: the mountains lie across this region from northeast to southwest, with some prominent peaks to the southeast including the summit of Mt. Nuluerhu (known locally as Daqingshan/Duanqinshan, or ‘Large Green Mountain’ in Chinese). Like Niuheliang, it is on a perennial branch of the Daling River (the Laohushan River or ‘Tiger Hill River’ in Chinese) that cuts through the mountains and is the main access route through the region. A systematic field survey of c. 70 km2 along the river valley has identified eight ceremonial sites with tombs or altars and another ten places where only pottery cylinder sherds were collected (IMAT and MAB 2005). The most common local bedrock is either red volcanic tuff or yellow sandstone, which provide appropriate stone for altar bases. More generally, this sacred area in the upper Laohushan valley again suggests the existence of exclusive Hongshan sacred places that were physically separated from everyday settlement and secular life (Fig. 12).
https://static-content.springer.com/image/art%3A10.1007%2Fs10963-013-9062-9/MediaObjects/10963_2013_9062_Fig12_HTML.gif
Fig. 12

Distribution map of the ceremonial sites at Sijiazi

The largest site at Sijiazi is called Caomaoshan (‘Straw Hat Hill’ in Chinese) and excavations there have uncovered an elaborate altar and tombs (Shao 2004). The tombs were built with stone coffins and the skeletons were placed with their heads to the east, and with jade objects as the main offerings. These tombs show features that are broadly similar to those found in the upper deposits period at Niuheliang, and therefore can tentatively be regarded as contemporary. The altar at Caomaoshan was built in as a three-layered, distinctively-shaped stone construction surrounded by an enclosure wall. Some commentators have suggested its shape deliberately evokes that of a tortoise (Shao 2004), which in later Chinese cosmology symbolised the earth and appears to be one of the most popular themes among Hongshan jade objects. Some female figurines made from red tuff were also found at this site, which may indicate that it served a similar function to the ‘Goddess Temple’ at Niuheliang.

The small sample at Sijiazi precludes any formal spatial analysis, but similar patterns to those at Niuheliang are arguably present here also. For example, there may well be a similar distinction between large and small sites within the overall Sijiazi complex and the eight known sites can arguably be grouped into four distinct groups, each with an altar and associated tombs. The largest and most complex site at Caomaoshan is located on one of the most prominent hilltops on the northern side with extensive surrounding views.

Dongshanzui

The ceremonial complex of Dongshanzui is located in the upper Daling River valley, just 30 km east of Niuheliang. It was the first such ceremonial landscape to be discovered and its excavation in the late 1970s had a great impact on Hongshan archaeological research (Guo and Zhang 1984). Recently, a systematic survey of a 200 km2 area around Dongshanzui has documented a landscape that includes at least four sub-groupings of ceremonial activity (Peterson et al. 2010; LPICRA et al. 2010). As a landscape, the area around Dongshanzui is similar to Niuheliang, with mountains stretching from northeast to southwest and the Daling River flowing through. The most view-prominent mountain peaks here are also in the southeast, especially one locally called Majiazishan (‘Horse Framed Mountain’ in Chinese), which lies on the opposite side of a wide river flood plain. As at Niuheliang, all sites at Dongshanzui were found on the gneiss bedrock units. The main monument found at Dongshanzui is a huge stone construction of complex design, which takes up almost the whole surface area of the site. It seems to represent a complex of altars in both square and round shapes that are symmetrically arranged along a north–south axis, surrounded by a stone enclosure wall and pottery cylinders. The sample is too small for us to be sure at this stage, but there might also be sub-distinctions among the ceremonial sites at Dongshanzui similar to the ones noted at Niuheliang, as only cylinder sherds and no monumental evidence have been found in some other localities. According to radiocarbon data, the stone altars in Dongshanzui site are contemporary with the larger sites at Niuheliang (Guo and Zhang 1984, p. 10), with a recalibrated date centred on c. 5700 BP (Fig. 7), and the site has also produced a lot of exotic artefacts comparable to those at Niuheliang, including animal-shaped jades, pottery vessels and pregnant female figurines.

Despite these similarities, a crucial difference between Dongshanzui and Niuheliang is that, at the former, these ceremonial activities are found in an area with settlement remains (although it is as yet unclear whether these are contemporary or earlier), and, currently, in the complete absence of tomb activity (Fig. 13). Although further research in the region may conceivably alter or nuance this picture, at present it appears that the ceremonial activities associated with the animal-shaped jade objects, pottery cylinders, female figurines and stone-clad altars not exclusively associated with burial practice, but under certain conditions could be deployed either for elite funerary or non-funerary rituals. Indeed, some tombs with painted pottery cylinders were also discovered at residential sites such as Fushan, while some stones circles were found on hilltops close to residential sites in the Chifeng survey area (Peterson et al. 2010, p. 5757; CICARP 2003, pp. 31, 113). Therefore, while a mixture of funerary and ceremonial activities at Niuheliang continues to stand out, we cannot assume it represents a ceremonial type site.
https://static-content.springer.com/image/art%3A10.1007%2Fs10963-013-9062-9/MediaObjects/10963_2013_9062_Fig13_HTML.gif
Fig. 13

Distribution map of the ceremonial sites at Dongshanzui

Discussion

The above discussion of Hongshan settlement and ceremonial landscapes has sought to offer context and wider insight into the dynamics of Hongshan society in general, particularly in its later phases. Ceremonial activities, and in some cases elite burial, focused on a few key sites, were clearly of great importance to large-scale integration of Hongshan society, especially given the evidence for very little hierarchical organisation in the settlement landscape itself. Although such sacred places are found in several different areas, they share many similarities in terms of their surrounding material and visual landscape. Most Hongshan ceremonial complexes found so far were preferentially located on the southwestern margins of the overall Hongshan cultural area (as we perceive it archaeologically), on the tops of low hills, within river valleys, and with high mountains as background, which led to their having expansive views of the surrounding countryside and potentially high levels of connection between individual locations. Manipulation of different kinds of local and/or exotic stone resource, for building and crafting, were clearly important, and we suspect that their on-site manipulation in terms of type and colour enabled micro-cosmographical expressions of perceived variability in the wider Hongshan landscape. Animal imagery is an important aspect of the portable material culture at these sites, with plausible but tentative identifications amongst the jade objects and figurines including owls, tortoises, dragons, bears, pigs, boars, hawks, fish, and silkworms (see also Guo 2010b). It is also not inconceivable that certain built structures, such as the altar complex at Caomaoshan, or landforms, such as Mt Mulan Shan at Niuheliang, may have been ascribed specific similar animal forms and meanings, but it remains difficult to prove this archaeologically.

Shamanic rituals and chiefly authority are the two kinds of influence most heavily emphasised by other commentators with respect to Hongshan ceremonial activity (e.g. Barnes and Guo 1996; Peterson et al. 2010). We would also add that there seems to be a spiralling dynamic of investment at at least some of these sites that involves: (a) their appearance, on present evidence, only in the later part of the Hongshan period at perhaps c. 5700 BP, and (b) the signs of spatial and chronological stratification within this ceremonial landscape, with the greatest levels of monumental and artefactual elaboration occurring in the suggested second phase of construction at Niuheliang. The quantitative analysis offered above also suggests that one feature of this latest stage, at Niuheliang at least, was the more concrete formalisation of the material relationships and landscape meanings associated with elite ritual activity, in particular, of spatially and contextually structured combinations of different materials (for example, clay and stone to name the two most archaeologically obvious), of both local and exotic provenance, for both fixed installations and portable material culture. In terms of location, the larger, later sites at Niuheliang also laid more emphasis on a structured internal arrangement, and possibly on intervisibility among localities and a position along a central axis of activity, with two prominent monuments, the ‘Goddess Temple’ and pyramid-shaped building at either end. It is also possible to argue that the substantially greater investment so far documented at Niuheliang itself, by comparison to other known ceremonial sites, suggests it was playing an increasingly important regional role. In any case, a plausible wider explanation for these increasingly formal landscape relationships is that they reflected and consolidated wider patterns of social and political stratification at this time, particularly in the Hongshan southwest.

The contrast with Hongshan settlement patterns is striking. Although both settlement and ceremonial sites can be seen to group into looser aggregations at several scales, and occasionally larger villages are found, there remain few if any signs of an organised settlement hierarchy. Moreover, some work on residential sites suggests both limited deposit build-up at any one locality and exploitation of a variety of landscapes (Teng 2010), at least raising the possibility that Hongshan residences shifted from time to time (though such a conclusion must remain tentative and we cannot as yet speculate about the possible tempo of any shifts if they did happen). In any case, under such circumstances of limited settlement hierarchy and possible non-stable residence, we can argue that, by the later phases of the Hongshan period, certain ceremonial sites were becoming the key integrative components that promoted the social, religious and political credentials of a regional elite.

Acknowledgments

This research has been funded both by ICCHA (International Centre for Chinese Heritage and Archaeology, based at University College London) and the Project ‘Formation and Early Development of Chinese Civilization from 3500 to 1500 BC: Research on Social Structure Reflected by Settlement Patterns’. We are also grateful to Dr. Xinwei Li for providing us with survey data from Sijiazi, and the three anonymous reviewers for their careful readings and valuable advice.

Copyright information

© Springer Science+Business Media New York 2013