XIONGNU TERRACE TOMBS AND THEIR INTERPRETATION AS ELITE BURIALS

Transcription

1 XIONGNU TERRACE TOMBS AND THEIR INTERPRETATION AS ELITE BURIALS Introduction 1 Terrace tombs appear at first glance to be quite homogeneous. They share a number of characteristics like tomb layout and construction, including a dromos on its southern side, a rich inventory displaying massive presence of foreign artefacts, such as chariots, imported metal vessels, foreign textiles, and numerous and rich horse gear to name only the most outstanding features. And all these tombs were re-opened and show a considerable amount of destruction, which is generally associated with looting. In the archaeological literature terrace tombs are named differently and terms like Elite, aristocratic or royal are frequently used indicating their interpretation 2. In some cases this interpretation is extended to the whole cemetery as being aristocratic or royal 3. But is it really appropriate to name both huge Gol Mod necropolises Elite cemeteries with each of around 400 (visible) graves in a distance of 60 km from one cemetery to the other? Is then the valley of kings of Xiongnu times located in Arkhangai aimag? The problem with these terms is the lack of definition and reflection in the archaeological literature 4. It is not the place or goal of this contribution to provide an extensive discussion of the terms but to compile and discuss the archaeological evidence on which further interpretation will have to rest on. The central problem lies in the fact that so far an archaeological analysis regarding status and rank of the males and females within Xiongnu communities, which would have to start with an analysis of the social differences in burial communities and end with a general scheme for defining rank and social stratification of those entities, is not fully established yet 5. 1 For their openness to discuss ideas and aspects of this contribution my special thanks goes to Jan Bemmann, Bonn, Bryan Miller, Philadelphia, and Ch. Erööl-Erdene, Ulaanbaatar. 2 See for example André 2007; Miniaev / Sakharovskaia 2007a; Konovalov 2008a; 2008b; Polos mak et al Miniaev / Sakharovskaia 2007a; Desroches / André this volume; Miller et al This becomes especially important since the meanings of terms like Elite, Aristocracy, Tsarskii, in German, French, English, and Russian, to name only some of the western languages, are not identical. 5 An analysis of burial customs of Xiongnu circular graves is provided by Törbat Furthermore an important step to researching the social structure of Xiongnu society was undertaken by Kradin et al where several differences in ranks of female and male graves were identified. Not included in their study are the rich burials since at that time the ostentatious burials from Tsaram and from Il movaia Pad were not researched and published yet. Honeychurch et al continued on this path utilizing in their study a regional approach. The authors identify a minimum of two social processes forming the status systems: a hereditary based status and within this a system of an achievement based status (Honeychurch et al. 2007, 141). They arrive at the conclusion that the majority of Xiongnu individuals in the lower Egiin Gol Valley were given very different mortuary treatment from those represented in the Burkhan Tolgoi cemetery (ibid. 143).

2 248 Of course, it is beyond any question that those monumental, richly furnished tombs like Tsaram, kurgan 7, T1 and T20 in Gol Mod or kurgan 20 from Noyon Uul are burials for the Elite, which can be used as a very neutral term for these burials and is interchangeable with ostentatious graves 6. As we are yet lacking the archaeological definition of Elite burials it is methodically not convincing to transfer terms from the Chinese written records, which are clearly describing the situation from a stranger s view and are hence in itself tendentious, in order to interpret the archaeological situation. It is even more difficult to classify what today is called a Xiongnu Elite burial. The question may be raised if all ramped terrace graves are burials for the Elite and for which part of the Elite? So far the group of ramped terrace tombs has not been studied in detail, which is the intention of this paper. As the database is rather thin some of the data is yet in the course of publication the ideas presented can only be preliminary. But there is a considerable amount of variation displayed in those tombs 7, which seems to be worthwhile taking a closer look at. Therefore I will first present the archaeological record and then take a brief look at the historical record suggesting ways of reading this cultural phenomenon of those squared tombs. Terrace Tombs in Gol Mod and beyond the Archaeological Record Until now around 20 terrace tombs have been excavated 8. It is not easy to compare the available data of them since this is hampered by three main factors: The first filter is the condition of preservation. While we have an abundant mass of textile as well as wooden artefacts from Noyon Uul, the sandy soil of Gol Mod for instance destroyed most organic material and only preserved it in small fragments under special circumstances. The second distortion results from the fact that the excavations of the square ramped tombs took place over a time span of more than 80 years which naturally results in different excavation techniques: with the exception of the Teploukhov kurgan all kurgans in Noyon Uul excavated in the course of the Kozlov expeditions were researched with a shaft method, which allowed only limited observations on the construction (Rudenko 1962). The same accounts for the tombs excavated by Navaan in Takhiltyn Khotgor between 1987 and 1990 (Navaan 1999). 6 Aristocratic is in my opinion too much loaded with meaning from Medieval European Society, signifying the governance of nobility. 7 See Erdenebaatar 2000; Konovalov 2008a, 35; 2008b, Cf. Konovalov 2008a, 5 et seq.; 2008b, 11 et seq. With the exception of Kozlov s expedition in Noyon Uul, which tombs are considering their time of publication fairly well published, we possess only few information from other tombs as those excavations have either taken place only in very recent years and hence are not yet published or in very early years and hence are published with only few information. Following tombs are included in the present study: Mongolia: eight kurgans from the Kozlov expedition in Noyon Uul (Rudenko 1969; Umehara 1960); Noyon Uul, kurgan 20 (Polos mak et al. 2008); Gol Mod T1 (André 2002; Desroches 2003; 2005; Mongolie 2003; André / Erööl-Erdene 2004) and T20 (Desroches / André 2007); Gol Mod 2: although Tomb 1 is not yet excavated, information on its size is available (Allard et al. 2002; Miller et al. 2006); Duurlig Nars, kurgan 2 (Eregzen et al. 2008); Takhiltyn Khotgor, kurgan No. 82 (Navaan 1999, tomb No. 1), kurgan No. 83 (Navaan 1999, tomb No. 2), kurgan No. 64 (Miller et al. 2008; Miller et al. this volume). In Russia: Transbaikalia: Tsaram, kurgan 7 (Miniaev /Sakharovskaia 2006a; 2007a; 2007b; 2007c); Il movaia Pad, Sudzha, kurgan 10 (Tal ko-gryntsevich 1999, 41; Pl. 6), kurgan 54 (Konovalov 2008a; 2008b); Tuva: Bai-Dag II (Nikolaev 2003 with older literature).

3 Xiongnu Terrace Tombs and their Interpretation as Elite Burials 249 The third difficulty lies in the different state of publication. Often older excavations are published with less information, whereas the results of the excavations conducted in recent years are mostly not fully published yet. Therefore few information is available from Gol Mod T20, Noyon Uul kurgan 20, Takhiltyn Khotgor K 64 or Duurlig Nars kurgan 2, while many articles by Miniaev and Sakharovskaia have been published on kurgan 7 from Tsaram 9. So far the only complete publication is the one by P. Konovalov on kurgan 54 in Il movaia Pad, Sudzha (Konovalov 2008a; 2008b). Despite these difficulties I think it is worthwile taking a look at this group of burials in order to advance in questions regarding social processes among the Xiongnu. Distribution and Layout of Cemeteries We find rampart tombs across a wide territory in nowadays Mongolia and Southern Siberia 10 (Fig. 1). Generally Xiongnu cemeteries can be divided in two different types (Miniaev 1985, 24): Type 1 consists of rectangular terrace burials and circular burials whereas type 2 is only made up of circular burials. In this study we shall focus only on burial places of type 1. The largest number of terrace tombs in one cemetery is so far known from Noyon Uul and from Arkhangai aimag, from the cemeteries of Gol Mod 1 and Gol Mod 2, which is also the centre of their distribution. As S. Miniaev pointed out, Xiongnu cemeteries are mostly located on Fig. 1. Distribution of Xiongnu terrace tombs in Mongolia and Southern Russia. 1 Takhiltyn-Khotgor; 2 Bai-Dag II; 3 Gol Mod 2; 4 Gol Mod 1; 5 Noyon Uul; 6 Selbe Gol; 7 Tsaram; 8 Il'movaia Pad', Sudzha; 9 Bor Bulag; 10 Duurlig Nars. 9 I will use the name Tsaram here instead of Tsaraam, which is mostly applied in English literature, since the correct transliterations should spell it this way. 10 We only focus on terrace burials, which can be clearly associated with Xiongnu archaeological culture, hence we exclude burials with stepped pits from China.

4 250 sloping terrain, remote from big rivers and clustered in several burial groups (Miniaev 1985, 22 et seq.). For nomadic communities of Central Asia locations of burial places coincide often with areas used for winter pastures, which play a central role for survival with all its social and historical implications (Hüttel 2001, 3 et seq.). Focusing on burial grounds with terrace tombs we recognize that the cemeteries can widely vary in their sizes from 20 to 1,750 ha (Tab. 1) 11. The three biggest ones are Noyon Uul and both cemeteries of Gol Mod in Arkhangai aimag 12. Cemeteries of the next largest group cover an area between 100 and 280 ha whereas the smallest cemetery of Takhiltyn Khotgor is only 19 ha big and located on the Western periphery. Name Length m Width m Area ha Noyon Uul (Süzhigt) Gol Mod Gol Mod Il'movaia Pad' Duurlig Nars Bor Bulag Takhiltyn Khotgor Bai-Dag II Tab. 1. Overview over area sizes of cemeteries with terrace tombs. Maps of Xiongnu cemeteries with terrace burials show that not only their layout differs but also the proportions of circular versus rectangular burials. In Gol Mod 1 burials are clustered in several groups (Desroches / André, this volume 316 Fig. 1). The dominating burial form is the rectangular burial with a dromos on the southern side. Of a total of 393 recorded burials (Desroches / André this volume 317), roughly half of them seem to be terrace graves. We can recognize that terrace tombs are often, but not always, accompanied by circular, satellite burials. Only around 60 km south-west of Gol Mod 1 the cemetery of Gol Mod 2 comprises again about 400 burials, which show clustering that is, however, not as distinct as in Gol Mod 1 (Fig. 2). It is interesting to note that on the south-eastern fringe of the cemetery two burial groups contain only circular burials. Both Gol Mod necropolises belong to the largest Xiongnu cemeteries known today in Mongolia. In Noyon Uul burials are also clustered in distinct groups (Fig. 3.1). The Mongolian-Tibetan expedition of Kozlov counted all in all 212 kurgans (Rudenko 1969, 12). A revised plan of the burial group of Süzhigt by N. Polos mak shows a smaller eastern cluster and a larger western cluster of burials (Fig. 3.2). Based on this map with its limited information, we count only The small cemetery from Bai Dag II is listed, however is not mentioned due to its extremely small size of 1 ha and its far off location on the periphery. 12 The large area of Noyon Uul has been roughly estimated by the small map published by Rudenko 1969 (cf. Fig. 3.1) encompassing the complete area of the different burial groups, despite the fact that between Süzhigt and Khuzhirt lies a seemingly empty area of about 4 km (cf. Fig. 3.1). The size of Süzhigt alone measures 25 ha (740 x 340 m).

5 Xiongnu Terrace Tombs and their Interpretation as Elite Burials 251 Fig. 2. Gol Mod 2, Arkhangai aimag. Map of the cemetery (Allard et al. 2002, 638). burials with a dromos on their southern side from a total of 79 identified burials. The portion of terrace tombs seems to be smaller than in the cemeteries of Arkhangai aimag. But considering the difficult visibility of tombs in the forest, this small number of terrace tombs might be misleading. In eastern Mongolia, in Khentii aimag, the two cemeteries of Duurlig Nars and Bor Bulag are with 100 and 60 ha smaller than the ones in Arkhangai aimag. As in Gol Mod 1 and 2 terrace tombs dominate (Fig. 4; 5). While we recognize two dense burial clusters in Duurlig Nars, Bor Bulag shows a much more even distribution. To the north, in Buriatiia, the necropolis of Sudzha in Il movaia Pad shows a different pattern with circular burials outbalancing terrace tombs by far (Fig. 6). Again the burials are clustered in several groups. Especially on the south-eastern fringe of the necropolis we note the exclusive existence of circular burials. On the western periphery of the Xiongnu Empire the smaller cemetery of Takhiltyn Khotgor consists also of terrace tombs and circular burials, arranged in groups (Miller et al. this volume, 301 Fig. 1). Terrace tombs and circular burials seem to be evenly balanced without a clear domination of any form. The largest terrace tombs in Takhiltyn Khotgor are in general smaller than the ones in Gol Mod 1 or Gol Mod 2. In Tuva, five square ramped burials are known from Bai-Dag II contrasted by 11 burials, which belong to the circular type (Fig. 7). While all terrace burials are attributed to the Xiongnu culture, some circular burials belong to the Skythian and Kokel culture, which gives

6 Fig. 3. Noyon Uul. 1 Overview over different burial groups (after Rudenko 1969, 122, Fig. 2); 2 Map of Süzhigt (Polos mak et al. 2008, 78, Fig. 2).

7 Xiongnu Terrace Tombs and their Interpretation as Elite Burials 253 Fig. 4. Duurlig Nars. Map of the cemetery (Gurvan Gol 1994). an insight in the relative chronology of that place (Nikolaev 2003, 262). The burials in this cemetery do not show any distinct clustering in separate groups 13. In order to understand the different layouts and clustering of burial groups we need to review the distribution of burials according to gender, age, social status and rank as well as the internal chronology and duration of usage of each cemetery. Of course these questions can only provide the methodological structure in order to derive any substantiated interpretation, as the available archaeological data does currently not allow to give an answer yet. 13 The cemetery of Selbe Gol seems to comprise only ten graves (Amartüvshin et al. 2009). I thank Erööl-Erdene for providing information on this burial group.

8 254 Fig. 5. Bor Bulag. Map of the cemetery (Gurvan Gol 1994). Fig. 6. Il'movaia Pad', Transbaikalia. Map of the cemetery (after Konovalov 2008b, Pl. 2).

9 Xiongnu Terrace Tombs and their Interpretation as Elite Burials 255 Fig. 7. Bai-Dag II, Tuva. Map of the cemetery (Nikolaev 2003, 262, Fig. 2). Sizes of Terrace Burials Although the general layout of terrace tombs is very similar, their sizes can differ considerably (Tab. 2). The largest tomb recorded so far is T1 from Gol Mod 2 with a terrace measuring 46 x 46 m (i. e. 2,116 m 2 ), whereas the sizes of the smallest one, located in Takhiltyn Khotgor measures 8.1 x 8.27 m comprising an area of only 66.9 m 2. Gol Mod 1 also shows great differences in burial sizes 14. In Gol Mod 2 the recorded 98 terrace tombs vary from 8 to 83 m in total length including rampart and terrace (Miller et al. 2006, 2). Also, circular burials show significant differences in their diameters: 250 documented small satellite burials stand in contrast to 85 large circular burials, which can reach a diameter of up to 28.3 m (ibid.) and thus are larger than small terrace tombs there. On the northern periphery this overlap is also visible in Bai-Dag II where two large terrace tombs are recorded (Fig. 7). Burial 9 measures an overall 14 Considering the fact that the measurements of those hundreds of terrace tombs from Gol Mod 1, Gol Mod 2 or Noyon Uul are not yet published, it is quite daring to state a medium size of the terrace tombs between 15 and 20 m (cf. Csíky 2007, 57).

10 256 length of about 30 m including the dromos. The three smaller ones (Nos. 16; 30; 32) are a bit more than 10 m long and thus are smaller than the large circular burials Nos. 3, 8, 24 and 29. The variation of sizes within the rectangular tombs (cf. also Fig. 8) as well as the considerable overlap of sizes between terrace and circular burial indicates that terrace burials despite their similar rectangular layout cannot be interpreted as one single homogeneous group. No. F Name Construction of burial 1 size of terrace dromos burial pit length (m) width (m) size (sqm) length (m) depth (m) 1 Noyon Uul, Verkhnii, kurgan 6 24,5 24,5 600,3 22,5 13,15 2 Gol Mod 1, T 20 30,0 30,0 900,0 21,0 18,3 3 Tsaram, kurgan 7 32,0 32,0 1024,0 20, Noyon Uul, kurgan 20 20,0 19,0 380,0 17,0 18,1 2 5 Gol Mod 1, T 1 30,0 30,0 900,0 35, Il movaia Pad kurgan 54 18,0 19,0 342,0 14,0 8,5 7 F? Takhilt, Navaan, T 82, unlooted 13,0 24,0 312,0 3, Duurlig Nars, T 2 9,0 11,0 99,0 15,0 8 9 Gol Mod 1, T 79 14,0 12,0 168,0 8,0 11 Takhilt, Navaan, T 83 12,0 16,0 192,0 4, F Takhilt, Miller, T 64 8,1 8,3 67,0 7,0 5,98 12 F Noyon Uul, Teploukhov, kurgan 24 14,0 15,0 210,0 12, Noyon Uul, Ballod kurgan 35,0 35,0 1225,0 11,3 14 Noyon Uul, Kondrat ev kurgan 26,0 26,0 676,0 17, Noyon Uul, Mokrii, kurgan 1 21,9 21,9 479,6 21,0 10,1 16 Noyon Uul, kurgan 25 18,4 18,4 338,6 18,0 8,76 17 Noyon Uul, Andreev kurgan 23,0 23,0 529, Noyon Uul, kurgan 23 18,4 18,4 338,6 17,5 10,2 19 Gol Mod 2, T 1 46,0 46,0 2116,0 37,0 Tab. 2. Table outlining dimensions of square tombs. The order of the tombs is the same as in Tab. 3, sorted according to the burial inventory. Grey: old excavations or insuffi cient information; F: female. Construction of the Terrace and the Burial Pit The construction of terrace tombs is roughly the same and varies only in detail: They are made of a rectangular, sometimes also slightly trapezoidal, soil terrace or platform, which is mostly elevated about 0.5 to 1.5 m above the ground. Some burials might simply have an enclosure without an elevated terrace as pictures of Duurlig Nars, tomb 2, suggest (Eregzen et al. 2008, 165, Fig. 3). The confirmation of this construction would need further information about the

11 Xiongnu Terrace Tombs and their Interpretation as Elite Burials 257 ancient surface, which has not been discussed so far. Usually the space of the burial construction is enclosed by dry masonry of one or more layers, often made of layers of horizontally placed stones 15. On its southern side, indicating the northern orientation of the burial itself, a rampart of rectangular or trapezoid shape is leading up towards the terrace, which is on the surface also enclosed by stones (Fig. 8). Once the rampart is excavated it appears that the dromos is leading downward to the central burial pit with the burial chamber. As far as we know, it generally does not reach the bottom of the burial pit but levels off higher up. As can be expected, the size of the terrace and the length of the dromos show a high correlation 16, meaning the bigger the terrace, the longer the dromos (Fig. 9.1). This indicates the symbolic value of the burial construction and the amount of invested labour in building such a huge burial. The enclosed terrace is often structured by stone rows on the surface, which divide the platform into different sections (Fig. 8). Variability is seen in the number of lines structuring the surface, which seems not to depend on the size of the mound. In Tsaram the surface of the huge mound appears to be without any stone lines and only a division by wooden beams is visible in the upper part of the burial pit, which is so far unique (Fig. 8.3). Mostly, a north southern stone line parts the mound as well as the rampart in a western and eastern half. The numbers of west east stone lines can vary to a great extend. While in Takhiltyn Khotgor 64 only one east west line structures the mound (Fig. 8.9), the over-grave structure of kurgan 54 from the Sudzha cemetery is divided by four such inner lines (Fig. 8.5). These partitions can either be built-up as stone lines only few layers high or sometimes be traced into depth and actually appear like walls as is the case in Il movaia Pad 54 (Konovalov 2008a, Fig. 8; 2008b, Pl. 8) or in Takhiltyn Khotgor 64 (cf. Miller et al. this volume, 302 et seq.). Most of the excavated burial pits are constructed like an inverse pyramid with steep steps, as documented in several cases 17. Geo-radar images from Gol Mod 1 show that a burial pit does not necessarily lead down in steps but can also be funnel-shaped (Fig. 10) which can be explained by the extremely loose sand there, in which the burial pit was embedded not allowing a construction with many steps compared to other locations. Often one or more stone layers are recorded in the burial pit. While for T20 in Gol Mod only one layer with stone rows was discovered at a depth of 8 m dividing the burial pit into several compartments, four real stone layers covered the burial pit in Tsaram 7 and even five layers in 15 Konovalov 2008a, Fig 19; 20; 2008b, Pl ; Desroches / André 2007, 56. According to the Russian description by Miniaev / Sakharovskaia 2007b, 130, also vertically placed stones are possible; misleading is Csíky s interpretation 2007, 57 of this passage stating that sometimes four orthostates flank the stone alignments referring to Tsaram. So far the published plan of kurgan 7 as well as the descriptions of its construction do not allow such a conclusion and it seems that the author s opinion is possibly caused by the English text (cf. Miniaev / Sakharovskaia 2007b, 130 with Miniaev / Sakharovskaia 2006a, 48); for clarification of construction details we need to wait for the final publication of kurgan 7. So far, the author knows only one single kurgan in a Xiongnu cemetery with four cornerstones: tomb 92 of Gol Mod 1 (cf. Haquet 2003, 106; Mongolie 2003, 202 Fig.). 16 We have to keep in mind that so far only data of about 19 terrace tombs is available which is rather few considering those hundreds of burials, whose measures are yet unpublished; this underlines that the presented ideas are preliminary. The high correlation (correlation coefficient between 0.7 and 0.9) between the length of the dromos and the size of the terrace appears to be quite stable. Even taking out outliers, like the extreme size of Gol Mod 2 T1 does not change the high correlation. The significance of the correlation is also not changed when basing the analysis on an area (in sqm) vs. length of its side relation to the length of the dromos. 17 Il movaia Pad, Sudzha, kurgan 54 (Konovalov 2008a, Fig. 10; 32; 2008b, Pl. 9.1; 23.2), kurgan 7 in Tsaram (Miniaev / Sakharovskaia 2007c, 44, Fig. 1) or Noyon Uul, kurgan 20 (Polos mak et al. 2008, 80 81, Fig. 5 10); Duurlig Nars, tomb 2 (Eregzen et al. 2008, 158) and Takhiltyn Khotgor 64 (Miller et al. this volume, 305).

12 258 Fig. 8. Tomb plans of terrace burials. 1 Gol Mod 2, burial 1 (Miller et al. 2006, 3, Fig. 1); 2 Gol Mod 1, T 20 (Desroches / André 2007, 58, Fig.); 3 Tsaram, kurgan 7 (Miniaev / Sakharovskaia 2006, 79, Fig. 1); 4 Noyon Uul, kurgan 20 (Polos mak et al. 2008, 79, Fig. 4); 5 Il'movaia Pad', Sudzha, kurgan 54 (Konovalov 2008a, Fig. 5); 6 Gol Mod 1, T 79 (Mongolie 2003, 117); 7 Takhiltyn Khotgor, burial 82 (Navaan 1999, Fig. 7-1); 8 Takhiltyn Khotgor, burial 83 (Navaan 1999, Fig. 1); 9 Takhiltyn Khotgor, burial 64 (Miller et al. 2008, 29, Fig. 4).

13 Xiongnu Terrace Tombs and their Interpretation as Elite Burials Fig. 9. Correlations of different factors. 1 Correlation of terrace length and dromos length; 2 correlation of terrace length and depth of the burial; 3 correlation of the length of the dromos versus depth of the burial.

14 260 Fig. 10. Geo-radar image of a terrace tomb in Gol Mod (Desroches / André 2007, 53, Fig.) Fig. 11. Noyon Uul, tomb 20. Stone layers covering the burial pit. 1 first stone layer; 2 second stone layer; 3 second and third stone layer; 4 fourth stone layer (1 4 Polos mak et al. 2008, 80, Fig. 6 7; 81, Fig. 8 9).

15 Xiongnu Terrace Tombs and their Interpretation as Elite Burials 261 Noyon Uul 20 (Fig. 11). So far in all excavated terrace tombs a thick stone packing above the burial chamber is reported 18. The depths, just as the sizes of the burials, vary greatly between 8 and 18 m (Tab. 2). It is interesting to note that we get a quite low correlation between the size of the terrace and the depth of the pit (Fig. 9.3): while T1 and T20 in Gol Mod 1 have terraces measuring 30 x 30 m, kurgan 20 in Noyon Uul with its upper, much smaller mound of 20 x 19 m has the same depth. The difference in ground soil does not seem to be influential as tombs constructed in pure sand (Gol Mod 1) are as deep as the ones built in harder ground (Noyon Uul 20). While we can argue that the sizes of the tombs and their dromos are in relation to each other, the depth depends on yet another factor. And it probably does not lie in its symbolic value for the highest Elite since we would then expect a huge tomb also being very deep. Possibly differences in time might be responsible for the depth of the tombs as outlined in the discussion below. Almost insignificant is the correlation between the depth of a terrace burial and the length of its dromos (Fig. 9.2). Even though keeping in mind the small number of cases, the low correlation 19 might nevertheless indicate that the function of the dromos is less to be seen in a pathway bringing down the dead body and more in its ritual significance of a procession way or something alike. Miller arrived at the same conclusion analyzing the burial construction of Takhiltyn Khotgor 64 where the dromos already stops at the first ledge while the burial pit continues down for another 4.4 m (Miller et al. 2008, 29). The same is the case in Noyon Uul 20 where the burial pit continued down for several more metres (Fig. 12). In connection with the construction of the burial ritual activities and depositions in various depths and places are recorded. As it is also the case for circular burials often a deposit of animal bones is found at the north of the deceased, regularly outside of the burial chamber (Fig. 13). The kind of animal can vary but as a rule only the heads and legs were deposited 20 : At the northern wall of the burial pit of T20 in Gol Mod remains of 16 horses and the skull of one ox on a thick charcoal layer were found 21. Similarly, in Il movaia Pad 54 skulls and leg bones of six horses, and possibly also of other animals, were discovered arranged in two rows (Konovalov 2008a, 26, Fig. 11; 2008b, Pl. 10). In Tsaram 7, at a depth of 11 m under the third stone cover, skulls, leg, and tailbones of horses, cows, sheep, and goats were laid out in one row (Miniaev / Sakharovskaia 2006a, 49). Skull and metapodials of another horse were found at the incline of the fifth step of the burial pit, directly north to the yoke of the chariot, in position to draw the chariot (idem 2007c, 46; 47 Fig.). But we find also deposits of artefacts in other places 22 : In Tsaram 7, for example, at a depth of 2.15 m, a deposit of ten mirror fragments was found in the centre of the burial pit below the 18 Miller this volume, 304; Miniaev / Sakharovskaia 2007a, The square root of r 2 is 0.49, indicating a low to medium correlation: a low correlation is between 0.2 and 0.5 while a medium correlation is between 0.5 and There is a noteworthy exception to this rule: in Gol Mod 2, burial 29, which is located to the immediate north of the terrace tomb. The burial marked by a circle of stones was well over a meter deep and contained the remains of a single horse, orientated to the north without the head and two vertebrae missing. Its connection with the burial complex is unquestioned (Miller et al. 2006, 10). 21 Personal communication of Hélène Martin, INRAP, Toulouse, who analyzes the animal remains of Gol Mod for the Mission Archéologique Française en Mongolie (MAFM). 22 Single artefacts were always found during excavations in the filling of the burial and can very often be attributed to looters activities, which tossed up things. So far only few examples can be named as well observed in situ situations, which most likely simply reflects the state of research and publication.

16 262 Fig. 12. Noyon Uul, tomb 20. Reconstruction of the burial with dromos (Polos mak et al. 2008, 86, Fig. 21). wooden beams structuring the pit 23. The mirror is not only fragmented but also incomplete. Nevertheless its size of 13 cm in diameter and its ornamentation can be reconstructed. The traces on the mirror surface indicate that the mirror had been broken by the means of heating and cooling it. Only on one central fragment traces of melting are visible, which is why this melting seems to be the result of a second step of heating since the other fragments were untouched by the heat (Miniaev / Sakharovskaia 2006b, 82; 2007c, 46). In T1 of Gol Mod a deposit was found at a depth of 10 m with a bronze basin and single bronze and iron elements of a chariot and bronze pieces belonging to the horse-gear (Mongolie 2003, 120 et seq.). In T20 of Gol Mod under the stone partition at a depth of 8 m single pieces of a chariot were deposited. The observations during the excavations in Sudzha 54, Il movaia Pad, show another facet and illustrate the complexity of the rituals conducted when erecting the tomb there: after the construction of the burial chambers they covered the area with pine cones, brush, branches and needles (Konovalov 2008a, 23; 2008b, 30). On top of it they erected a log mound, placed the animal sacrifices at the northern end of the burial pit and then spread across the whole 23 Miniaev / Sakharovskaia 2006b, 79 et seq. with Fig. 1; 2007a, 160; 2007c.

17 Xiongnu Terrace Tombs and their Interpretation as Elite Burials 263 burial pit a soft layer of pine cones and brush, on top of which a thin layer of earth was spread. Then a huge fire was lit over the burial chamber. When the flames were dying down the pit was filled. Possibly the tossed in charcoals come from other fires around the pit, which might have accompanied the burial ceremony (ibid.). A thick, dense layer of charcoal and burnt earth was also discovered at the centre of the mound of tomb 64 in Takhiltyn Khotgor (Miller et al. 2008, 29). Since this part was untouched by looting activities this also points to rituals conducted in course of the erection of the tomb. For the construction of the burial chamber itself it is interesting to note, that the ones of the monumental terrace tombs consist of three components, an outer and inner chamber in which the coffin was placed in the middle or slightly moved out of the centre toward the eastern or western side 24. Smaller terrace tombs only feature two elements either a grave chamber and one coffin like in Takhiltyn Khotgor 64 or it seems that there are only two grave chambers without a coffin like possibly in Gol Mod T79 or the Ballod kurgan of Noyon Uul. In sum we can state that reviewing the construction details of the terrace tombs reveals a large variability among these graves. The same is the case when we look at the inventory. Inventory In the overview we see that terrace tombs show quite a broad variability in the quality of their furnishing (Tab. 3). Despite the fact that all listed burials were heavily destroyed, so that none of the inventory can be regarded as complete 25, we can distinguish so far two different groups of terrace burials: the first group appears to be homogeneous in many respects, even though the information will have to be completed by the full publications of Tsaram 7, Gol Mod T 20 and Noyon Uul 20: All terrace tombs of the first group seem to be furnished with a Chinese chariot. Mostly different sets of horse gear accompany the dead, some of them are very prestigious encompassing silver foil or gildening (André 2007, 73). They generally include round and elongated phalerae (Erööl-Erdene 2008, 144 Chart 1). Chinese lacquer as well as Chinese imported metal vessels are normally associated. Bronze cauldrons appear often as well as bronze sticks, which are interpreted as handles of the coffin 26. Nephrit stones and bronze mirrors round off the picture. Generally the coffin is decorated with gold foil and quatrefoil decoration in the ever repeating pattern of lattice work. Another feature observed in terrace tombs is the fact that many, if not all things, were covered with or wrapped in cloth. 24 Coffins not in the centre but moved to the eastern side are known from Noyon Uul 1 and 6; coffins in the centre: Noyon Uul 23, 24; Gol Mod T1 and T20; Tsaram An exception to this rule is Takhiltyn Khotgor 82 (Navaan 1999). 26 Miniaev / Sakharovskaia 2007c, 53; the same is also reported from Duurlig Nars, Tomb 2, for both cases the full publication of the feature will give clarification in this question. Strange remains in this context burial 82 of Takhiltyn Khotgor, which was unlooted but was nevertheless only equipped with two sticks, both deposited on the eastern side, one placed at the northern, the other one at the southern end of the chamber (Navaan 1999, Fig ). Before its interpretation as handles, the idea was proposed that these iron-copper-sticks resemble ceremonial sticks or scepters (Desroches 2003, 1178; André / Erööl- Erdene 2004; André 2007, 74).

18 264 No. F Horse gear Vessels Other inventory Date Tab. 3. Overview over the most important artefacts of the inventory in terrace tombs. Grey: old excavations; F: female. If we take a look at tomb 64 or 82 from Takhiltyn Khotgor and T 79 from Gol Mod 1, it seems that the difference of the inventory cannot only be ascribed to the fact of the looting. All tombs of the second group are smaller and less deep than the ones of the first group, none of them is equipped with horse gear of precious metal (silver), which seems to be the most outstanding characteristic of the first group. Possibly the same is valid for Chinese metal vessels. This leads to the question about the differences or maybe better overlap between the second group of terrace tombs and circular burials on the basis of their inventory. We know for instance mirrors (cf. Filippova 2000), cauldrons, lacquered bowls as well as lattice-quatrefoil

19 Xiongnu Terrace Tombs and their Interpretation as Elite Burials 265 decoration from a number of circular burials 27. While differences within circular burials are identified (Honeychurch et al. 2007), this has not yet been applied on a broader basis, including also square ramped tombs. It is interesting to note what in all terrace graves is rarely included: As far as we can tell weaponry does not play a significant role among the grave goods, since bows and arrows or swords are mostly missing 28. Can this be interpreted to that effect that it was not very important to display the role of the deceased as a warrior, or that the abdication of weaponry exactly displays the status and rank of the Elite beyond the rank of the warrior? Or were weapons simply the first thing to be completely removed in the process of destroying or looting the tomb? The topography of the burial inventory can so far hardly be reconstructed considering the lack of profound information from the Noyon Uul burials 29 or of not yet published information from Gol Mod T20, Noyon Uul 20 and Tsaram 7, although already a good deal of description for the latter is available (Miniaev / Sakharovskaia 2007c). The same accounts for tomb 2 of Duurlig Nars (Xiongnu Tombs 2009). The information about the arrangement of the goods in the burial chambers is necessary in order to better understand the differences and similarities of these terrace tombs in the future. So far we can barely outline and suggest preliminary trends, which need to be studied upon availability of further data. It seems that big storage vessels are to be deposited in the northern part of the burial chamber, as is the case in Gol Mod T20 and Il movaia Pad 54 (Fig. 13.1,2) but this is also reported for tomb 2 of Duurlig Nars (Xiongnu Tombs 2009) as well as Noyon Uul, kurgan 1 and 25 (Rudenko 1962, 108; 113). Possibly lacquered vessels are found north of the coffin or at least in the northern part of the burial, as we know from Gol Mod T20 and also from Noyon Uul, kurgan 23 and 24 (ibid. 112 et seq.). In Takhiltyn Khotgor 64 a red painted tray was also placed north of the coffin (Fig. 13.4). Horse gear often belongs to the eastern side, which is the case in Gol Mod T20, Il movaia Pad 54, Takhiltyn Khotgor 82 and also in Noyon Uul 25 (ibid. 113). In Tsaram 7 horse gear comes from the eastern and western outer corridor (Miniaev / Sakharovskaia 2007c, 50 et seq.), while in Duurlig Nars, tomb 2 horse gear was deposited in the east and south. It is interesting to note that prestigious elements of the horse harness are found in Tsaram 7 and Gol Mod T20 in the eastern part of the burial (Fig. 13.1,3). So far no clear picture is available about the deposition of metal vessels, be it of local origin like the cauldrons or foreign Chinese vessels. Directly north of the coffin we know two cases where the golden sun and moon were found (Fig. 13.1,4) Cauldrons are for example known from Egiin Gol, Burkhan Tolgoi, burial 63 (Törbat et al. 2003, 225 Fig.), 73 (ibid. 235 Fig.); Ivolga, burial 119 (Davydova 1996, Pl. 35.4) or Dyrestui kurgan 43 (Miniaev 1998, Pl. 21.6) to name only few; lacquer bowls are also known from circular graves, e. g. from burial 50 of Il movaia Pad (Konovalov 1976, 49 Fig. 22) and for quatrefoil decoration manufactured from different materials, like iron or birch-bark, see Mongolie 2003, 172 Fig. or Miller et al. 2008, Exceptional is Tsaram 7, where in the southern part of the eastern internal corridor a lacquered wooden quiver with iron arrowheads was found (Miniaev / Sakharovskaia 2007c, 53); in the south part of the coffin a ritual sword was discovered. In other graves only very few pieces of weaponry were found, like a single arrow head in Gol Mod T 20 or fragments of bow strengtheners from Il movaia Pad 54 (Konovalov 2008a, Fig. 48,4; 2008b, Pl. 37.1). 29 Having only the publication of Noyon Uul Russian and German version which differ sometimes in the description of the inventory localisation at hand and not any archive material, no well documented information about the topography of the burial inventory is available for kurgan 1, 6, 23, 24, 25, the Ballod and Andreev kurgan (Rudenko 1962, ; 1969, ). 30 This northern position for sun and moon is also known from circular burials, where they are made of iron, like in Burkhan Tolgoi, grave No. 1 (Törbat et al. 2003, 173).

20 266 animal sacrifice lacquer vessel (or imitation) metal vessels sun and moon prestigious horse gear horse gear storage vessel Fig. 13. Schematic grave plans with distribution of elements of the burial inventory (all burials orientated to the north). 1 Gol Mod T 20; 2 Il movaia Pad, kurgan 54; 3 Tsaram, kurgan 7; 4 Takhiltyn Khotgor, kurgan 64; 5 Takhiltyn Khotgor, kurgan 82. Even though the data is still quite fuzzy, there can be seen some underlying rules of placement, indicating similar beliefs. Especially among the first group uniformity of grave layout and grave furnishing becomes apparent. This uniformity displays the interconnection of the leading group of people buried in those graves as well as the communities who share the same beliefs and value system including status symbols. While generally in literature this first group is associated with the highest rank, most researchers think of a possible interpretation of the burials as one of a chanyu (e. g. Miniaev / Sakharovskaia 2002, 112; Desroches 2003, 1178 et seq.). But the question needs to be raised how we can interpret the second group of square ramped burials. So far there is not enough data to prove that the second group could

21 Xiongnu Terrace Tombs and their Interpretation as Elite Burials 267 either be connected with female burials 31 or interpreted as a chronological phenomenon (see below). And of course with the few skeletal remains it will also be very difficult to research age differences. Since we barely can exclude one of these possibilities, the assumption that the second group possibly reflects elite burials of a lesser status or rank stands on somewhat shaky ground. The Destruction in the Burial Another feature uniting all rectangular terrace tombs is the fact that they were reopened, a process generally referred to as looting, thus implying the intention 32. Usually one robber s shaft is clearly visible, but in Tsaram even two of them were identified (Miniaev / Sakharovskaia 2007a, 161). The results in Duurlig Nars 2 are spectacular, where traces of a cage construction were documented, which give a first insight in the robbing process (Eregzen et al. 2008, 157 et seq.; 165 Fig. 5; 6). Even though we do not have data yet to determine the point of time when those tombs were reopened it is reasonable to argue that this process did not take place very long after the burial ceremony since the burial chamber must still have stood when the opening had taken place in order to remove objects. In Gol Mod T20 for instance the lid of the coffin seemed to have been pushed aside, which could have only been done in an intact burial chamber 33. In my opinion it is far from evident to identify whether we see here the results of a looting process 34 or of a destruction process, which is not connected with grave robbery, and I question that the term looting thus determining the intention does correctly describe the phenomenon visible. The result we see in the burials is the fragmentation of objects on the one hand and we see the incompleteness of an object or a group of artefacts on the other hand. Both are the result of a multiple-step process. We can either assume that an object is destroyed and partially deposited or that an object is deposited, then destroyed and afterwards pieces were removed, or a mixture of both. So the question to be raised in order to approach this phenomenon is whether we can provide an informative basis for either of those ways. For the depositions above the burial chamber the placement of fragments is part of the ritual. Fragmentation is a main feature of the deposit in Gol Mod T1, where only parts of a bronze basin were found as well as single pieces of horse gear or chariot elements, maybe as pars pro toto. And as pointed out above in the case of Tsaram 7, parts of the mirror were deposited already in pieces (Miniaev / Sakharovskaia 2007c, 46). In T20 of Gol Mod it seems also that the Chinese chariot was deposited disjointed (André 2007, 74) and in T1 only elements of a chariot were recovered (André 2003). For these cases of depositions we can exclude fragmentation or incompleteness being the result of looters activities. 31 But indeed both graves, the un-looted burial of Takhiltyn Khotgor 82 as well as the recently excavated burial 64 are both graves of females (Miller et al. 2008, 29 and in this volume 304). 32 For example Miniaev 1985, 24; Miniaev / Sakharovskaia 2007c, 46; Csíky 2007, 58; Polos mak et al. 2008, 80; Desroches 2003, 112; Miller et al. this volume, 307 et seq.). 33 The same is also suggested by the excavations in Gol Mod 2, where in a satellite burial a complete horse head was tossed up and not only single parts of it; Bryan Miller, personal communication. 34 With the term looting I mean the process of grave robbing, which aims at taking either the corpse or his belongings or grave goods.

22 268 As for the inventory of the burial chambers it is more complicated since this part was reopened as the shafts leading to the roofs of the burial chambers indicate and it is impossible to determine whether incompleteness is solely the result of looting or whether partial deposition was also a component of the burial rite. In order to decide this as well as the question how much and which elements were taken out of the burial we would need an undestroyed monumental tomb. So far there is only one terrace tomb which has not been reopened: Takhiltyn Khotgor 82 (Navaan 1999), which also does not help answering the question since first of all it is situated on the western periphery, second, it is a comparatively small terrace burial, third, it is the burial of a woman and fourth, its inventory cannot be compared with those monumental tombs of the first group. In Gol Mod T20 the amount of destruction was visible everywhere, to name only some: the quatrefoil coffin decorations were in most cases ripped off leaving only traces of iron on the coffin lid and the bodies of all metal vessels, none of them complete, were broken with their single pieces scattered in the eastern corridor. Destruction seems to be prevailing in T20, while still a lot of artefacts, also prestigious ones of different functional groups, like metal vessels, horse gear or jewellery remained in the grave 35. This might speak against grave robbing as an act of personal gain. Before continuing further with the interpretation we need to understand the meaning of those burials. While most questions remain unsolved here, we probably gain a better picture of these processes once the monumental tombs excavated and well documented in the recent years are fully published. Dating Ramped Terrace Burials This applies also to the dating of terrace burials. So far only Noyon Uul, kurgan 6, Tsaram, kurgan 7, Gol Mod T1 and Il movaia Pad, can be regarded as well dated. In Noyon Uul, Tomb 6 was dated with regard to the inscription on the lacquer bowl, which was produced in the second year BC, providing a terminus post quem. Kurgan 7 of Tsaram was dated by the inscription on a lacquer vessel, which was probably produced between 8 BC and 4 AD 36 and also by radiocarbon dates obtained from the kurgan 7 as well as some surrounding satellite burials which according to Miniaev and Sakharovskaia point also to the 1st century AD We know also objects which were deposited already «broken» or rather incomplete: in the unlooted burial 82 from Takhiltyn Khotgor a complete copper basin with two handles was part of the inventory but it is remarkable that nevertheless a small portion of the rim is missing, which might also be due to the long use. However, the same is the case in Gol Mod T20 where except for a portion of the rim a complete pottery vessel stood beside the coffin. Thinking of the Chinese mirrors of which were always only fragments found, Filippova interprets this as a result of the burial rite (cf. Filippova 2000, ). Possibly also for other artefact groups this incompleteness war ritually required. 36 Pirazzoli-t Serstevens 2007, 58 pointing out that this date only gives a terminus post quem. 37 While agreeing with Miniaev that the burial complex, i. e. central kurgan and satellite tombs, were erected in a short period of time unfortunately his calibrated 14 C-Dates (OxCal) fall into a broad range between ca. 200 BC to 130 AD and therefore do not reflect this. This is why a broader and critical discussion of the radiocarbon dates themselves (sample size, material, place of sample) by the authors would have been appreciated before they conclude that the calibrated radiocarbon dates support a date between 30 and 120 AD (Miniaev / Sakharovskaia 2007c, 54; cf. also Konovalov / Brosseder 2008, 54 fn. 3); the youngest radiocarbon date out of the whole complex is a sample taken from the Chinese Chariot which is dated between 50 and 130 AD (Miniaev / Sakharovskaia 2007c, 54).

23 Xiongnu Terrace Tombs and their Interpretation as Elite Burials 269 Fig. 14. Il'movaia Pad', Sudzha 54. Radiocarbon date of one animal bone (Konovalov / Brosseder 2008, 53, Fig. 1). For T1 of Gol Mod a radiocarbon date was taken dating the burial between 20 and 50 AD but also the analysis of the chariot and mirror point to the same time period (André 2003, 124 et seq.; Mongolie 2003, 120 et seq.). Assuming that central kurgan and satellite burials were erected within a short time span we can also date the monumental Tomb 1 complex at Gol Mod 2 with a burial arc of 27 satellite tombs to mid 1st century AD 38. A new radiocarbon date is also available for Il movaia Pad 54 39, which dates the tomb between Cal BC 30 to Cal AD 60 (Fig. 14). We can therefore conclude that all excavated monumental terrace tombs, which can be dated so far, have been constructed at the earliest around the turn of the era, most probably in the 1st century AD, possibly in its first half or in the middle 40. But this can only be assumed for the monumental terrace tombs of the first group since so far none of the smaller ones are dated in terms of absolute chronology. Considering the fact that possibly within this group of dated terrace tombs the one from Il movaia Pad 54 is a bit smaller and less deep while its dating might possibly point to a somewhat earlier date than the other monumental terrace tombs, we 38 In burial 20 a TLV mirror probably from the Wang Mang period (9 23 AD) or soon thereafter was found (Miller et al. 2006, 13). A radiocarbon date for this tomb supports this date (cf. Miller 2009). 39 This dating was made possible by a grant from the Gerda Henkel Foundation in the frame of a project, which aims at establishing a finer internal chronology for the material culture attributed to the Xiongnu. The radiocarbon analysis was done in BETA Laboratories, USA. 40 One could argue that so far not enough terrace burials are excavated to come to such a conclusion, yet I do not think it is relevant here since the argument can be turned around and one can ask: why are all excavated and dated terrace burials (so far) not earlier than 1st century AD?