Ērān ud Anērān
Webfestschrift Marshak 2003


Archeology of Funeral Rituals as revealed by Tibetan tombs of the 8th to 9th century

Amy Heller*


This paper discusses artefacts in metal and bone recovered during archeological investigations of Tibetan tombs of the 8th to 9th century in Qinghai province. These artefacts are compared with the descriptions of Tibetan burial rituals of the period according to Tibetan manuscripts from Dunhuang, now conserved in the Bibliothèque Nationale, Paris. These artefacts correspond in many respects to the descriptions of Tibetan burial rituals of this period. Their function reveals ancient religious practices which are a synthesis of Buddhist and non-Buddhist rituals then prevalent in Tibet.

Recent archeological investigations in Dulan county, Qinghai province have discovered numerous 8th to 9th century tombs which contained religious artefacts as well as remains of skeletons, clothing, articles of daily life. It is my hope that those who study the modern period may see some contemporary parallels with these ancient burial practices. The Dulan tombs were investigated by the Chinese archeologist Xu Xinguo and his team, who believe that the people buried may be Tibetan or possibly 'A zha.1 Whatever their ethnic group may have been, Professor Xu acknowledges that the system of burial followed Tibetan customs. The correspondence with the architecture of Tibetan tombs in central Tibet is clearly evident in this photograph of the largest tomb (M1) prior to archeological investigation (plate 1). The spatial organization of the main tomb and ancillary tombs also follows known Tibetan models.2 A brief review of the spatial organization of the Dulan M1 tomb complex will help to consider how the archeological investigations both confirm and refute the data given by texts of rituals of Tibetan burial from Dunhuang manuscripts, earlier translated by Professor Stein and Mlle Lalou, in particular. My discussion will then concentrate on two objects recovered during the Dulan excavations: a silver reliquary to hold sharira , ring srel, objects revered as bodily remains of the Buddha, and a linga drawn on bone. These two objects seem to me to exemplify the artefacts of religious practices indicative of simultaneous coexistence of Buddhist cult and non-Buddhist Tibetan burial rituals.

The photograph of Dulan M1 (plate 1), which is the main tomb at Reshui, «warm water valley» shows the trapezoid shaped tomb ca. 1985, before any investigation. Total height: 47 meters, 160 meters long, 60 meters wide. Although this appears to be a two level construction, on the top of the entire mound were found traces of a foundation of a small chamber, destroyed long ago. What was then the upper level measured 12 meters high, and contained a rectangular chamber about 6 meters long, 5 wide, 2.25 ceiling height , walls one meter thick. This chamber contained bones of many animals, primarily sheep, horses, yak and deer.3 This entire upper level was virtually totally excavated in order to reach the lower level, which also had a chamber as we will see. The mountain behind the tomb is north, while to the south lies a river valley. In plate 2, we see the M1 tomb as excavated when I visited in 1997: the upper layer missing due to excavation which revealed a cruciform shaped chamber at the center of the lower level of the trapezoid.4

Although the tomb had been robbed long ago, in the four sections were found grains, wood, vast quantities of textiles, some garments, some banners, utensils for eating , and wooden slips byang bu which provide inventory of contents written in Tibetan dbu-med script. Although partially effaced, it is still possible to read words such as btsan po (sovereign), pho brang (palace or prince) and dar silks of different colors. However, today these chambers revealed no human remains, no corpse, nor any effigy of a human body.5

This tomb of packed earth and cut stone reinforced by log beams was constructed as two trapezoids, the smaller trapezoid built on top of the other.6 This trapezoid shape is likened in the literature to a mountain ri, or a tent gur. 7 The shape is distinct from what may be a construed as a rectangular tower, khar /mkhar, which shape is described by lDe'u chos 'byung for tombs only one generation before Srong btsan sgam po.8 Srong btsan sgam po's tomb and all his descendents' tombs in the royal necropolis in central Tibet follow the trapezoid shape but above all, the difference lies in the interior of the trapezoid where there is construction of the cruciform chambers, which is believed to be characteristic of the Tibetan tombs of this period.

However, the M1 tomb is not isolated. It is constructed within a complex of numerous tombs, both above and below ground, as evidenced by the following discussion. In this context of study of Tibetan conceptions of space, this Dulan-Reshui M1 burial complex raises questions about the implications of underground and above ground space.

The principal tomb is at center, constructed above ground, gigantic in comparison to the others which suround it. No less than 9 above-ground tombs, all trapezoidal in shape. These tombs are all relatively the same height and dimension. They are now empty, robbed long ago. However, immediately in front of main tomb, five lines - these are trenches for below-ground burial and at at the sides of the trenches, circles indicate circular pits also for below-ground burial. The five trenches, 3 meters apart , running east to west reveal the spectacular discovery of 87 complete skeletons of horses. It is pertinent here to recall the Dunhuang chronicle describing one hundred horses sacrificed on the tomb of blonpo mGar, principal minister of the Tibetan government, but this was, if I may say so, literary description of history, and perhaps literary hyperbole.9 On the other hand, 87 complete skeletons of horses is simple archeological accuracy. The description of Prof. Xu Xinguo is as follows:

" the remains of 87 sacrificed horses, a large number rarely encountered elsewhere, were found in the trenches: no. 1 trench contained 16 horses, no. 2 17, and number 3, 19 horses. the skeletons were arranged in different postures: some lay on their sides, some lay supine and others prostrate. A large boulder had been placed on each horse. From this we can assume that the horses were alive when they entered the trenches. In the center of the main no.1 burial trench was a gold and silver sharira reliquary casket, on top of which (yet another) large rock had been placed."10

The circular pits beside the trenches also contained animal remains underground, covered with a 50 cm thick layer of loess and a layer of topsoil. In total there were 14 sacrificial pits on the east, and 13 on the west. At the front rows, towards the north, there were hooves of sacrificed cattle and one skull per pit, while to the south, there were four pits containing complete dog skeletons. Two pits only contained large stones, no animal remains.

In the northeast, below the tumulus and inside the exploratory trench between the mountain and the tumulus, another circular pit, 1.2 m in diameter and 40 cm in depth, held 27 skulls of sheep and yaks.

To summarize the Dulan necropolis at Reshui M1, there are numerous underground cavities for sacrifice of animals or for the dismembered remains of animal bodies, slaughtered before burial. Perhaps approximately 200 animals were killed. The main tomb had an upper chamber for ritual above the tomb, a chamber for animal remains within the first level of the tomb, and a lower level with the cruciform chamber apparently for offerings. Was the corpse interred in the center of the cruciform chamber pillaged long ago? The secondary mounds beside the main mound are all empty, but believed to be of similar time and construction as the main mound. There is no indication of human burial here, neither human sacrifice or accompaniment of the mound. Yet these constructions do seem to be tombs, When confronted with textual citations of Tibetan burial rituals, or historic annals accounts of these rituals, Haarh's understanding had been that " the royal tomb spur khang, that is the funeral chamber and the corpse was in principle sub-terranean, below earth, directly dug into the earth, or covered by a mound of earth." 11 The Dulan burial complex indicates that the funeral chambers were not sub-terranean constructions, but clearly defined tombs of monumental above- ground architecture. In the literary and historic accounts, the yak, the mdzo, the sheep and the horse are the divine animals. In the description of Dulan tomb M1, the sheep seem almost to be lacking while they are strongly emphasized in the rituals and historic literature. Professor Stein, and Erik Haarh too, differentiated their functions, which I think appropriate to recall here: the horse was the mount of the dead, the guide through the mysteries, while the sheep had the task of finding the path to the dGa' ba'i yul, the land of Joy, the yak guides the demons in the rituals, thus he/she if mdzo has the task of separating the dead person from any/all demons on the path. Thus the yak has a protective function. Yet in the rituals there is another creature - the dbon lob, with most enigmatic roles, identified by Professor Stein as humans. In Dulan- Reshui M1, as I mentioned earlier, today there are no human bones, not anywhere - not in the tomb, not in any sacrificial trench. But other excavations by Professor Xu and his teams have clearly revealed human sacrifice. 12 This corresponds to citations from Chinese historic sources but it also raises questions - were the humans part of the burial ceremony but not killed? 13 Were they the custodians, guardians of the tomb and its contents? What to think of a tomb empty of all human occupants?

In terms of the Tibetan literary models, if we review the Dunhuang manuscripts studied by Professors Lalou and Stein, it is rapidly apparent that the texts were re-organized in much confusion, probably due to the copyist as suggested by Stein.14 Whatever disorder has come to PT 239 and 1042, it is clear that there are several phases of sacrifices, a procession of animals and offerings. This would seem to correspond to the live burial in trenches of the horses and the offerings of selected parts of sacrificed animals in pits. Mlle Lalou remarked the dismemberment of an animal into 13 parts, 12 members plus the head ,which is confirmed by the arrangement of hooves and heads in some of the circular pits - the animal was beheaded, then quartered, then the legs were cut in two, leaving the lower extremity of the hoof as seen in the photograph. 15 For instructions such as avoiding sacrifice of the horse of such and such color, the archeological remains do not allow us to determine the color of the horse, for we are left only with his skeleton. Many of the rituals' instructions in fact cannot be archeologically verified.

However, readings of Haarh of numerous historical documents led him to reconstruct the tomb of Stong btsan in 'Phyong rgyas in central Tibet, as having a central cruciform chamber which is in great correspondence to the actual spatial configuration of the Dulan M1 main tumulus. Although none of the 'Phyong rgyas tombs had horse burial trenches, the Tibetan tomb complexes at Khrom chen near Lha rtse and Ji du'i in central Tibet, north bank of gTsang po east of bSam yas, indicate the correspondence of the spatial organization of tumuli burial combined with burial trenches primarily for horses in several regions of Tibet during the sPu rgyal dynastic period.16 Lion statues in stone and rdo-ring stone stele are also a common element among these burial complex, although not described in the literature. Although I have had no access to the archeological statistics, another common factor among these tombs would appear to be the combination of above-ground burial constructions for humans and below-ground burial for animals, which may well have implications of religious and/or hierarchical nature. Yet, a possible correspondence to the traditional tripartite division of space in Tibetan ritual and literary accounts may be a different type of tomb - rather than constuction in a valley, such as the main necropolis at Dulan-Reshui, construction at elevated levels. Further west in the same river valley of Dulan M1 is a big hill, and in the midst of this desert-like plain, this hill is almost a small mountain. Plate 3 shows a photograph of the ridge of this hill where one may see the trapezoid construction of a tomb which has not yet been fully excavated, but is estimated by archeologists to measure approximately 12 meters by 15 meters, at least a height 6 meters. Does such a construction represent -even more than the tumuli constructed in the flat plain - the idea of construction in a space higher than the ground in a tripartite concept of the division of underground 'og, the earth sa and sky gnam?17. Or is such a construction simply analogous to the sacred mountain linked to the ancestors of the sovereign btsan po? Is the division of space only two-fold - underground and above ground- and could this possibly reflect a more archaic concept of territory as Haarh had written:

The ancient pre-Buddhist and pre-Bon concept of existence seems to have comprised two worlds of existence, that of man or Sa, the earth, and that of the defunct of 'Og, the underworld. The idea of heaven (gnam) in the sense of a third and upper sphere of the world may be a later development, which is somehow connected with the rise and spread of the (organized ) Bon religion in which a particular signicance and importance is ascribed to Heaven or gNam. The idea of the world as a Ga'u, the closed space of gNam-Sa, seems to represent an intermediate stage of development towards the idea of world of three spheres.

What differentiates the findings at Dulan M1? Although one must be cautious due to the robbery of the tomb in past times, the reliquary seems to me to be one factor indicative of simultaneous respect and/or syncretic practices of Buddhist and non-Buddhist rituals. The prominence of this object , buried in the center of the front trench amidst live horses, clearly signifies its importance. The reliquary consisted of a wooden container covered with sheets of gold plated silver. It was buried under a boulder, just like the horses, in fact the pressure of the rock caused the object to disintegrate during excavation, revealing a bed of ash, indicative that the container had once been burned. Or had it served to contain a fragrant burnt offering as well as a bone relic ? From the fragments (plate 4/5)now in Xining Archeological Institute, Professor Xu has reconstructed the shape. In comparison to this model (plate 6) of a reliquary from the Tang temple of Famen si, consecrated in 874A.D., the shape of the object is indeed quite similar, but the Chinese reliquary has no wood beneath the metal plates. The Dulan reliquary is distinctive also for the choice of the species of birds, carved out in silver and then partially gilt, is distinct. The side panel has the largest decoration - honeysuckle buds and vines surround a bird which is an imaginary hybrid - is it a standing phoenix against honeysuckle vines? Or do we see claws of an eagle, body of a pheasant, head of a female pheonix, while the sides of the box have the same flower and vines, but small iron birds with wooden heads with incrusted turquoise eyes.18 Tang reliquaries typically have a series of containers, gold inside and silver outside, approximate dimensions 10 cm long. They are entirely made of metal, usually carved repoussé with Buddhist emblems, while the reliquary excavated from Dulan-Reshui M1 was constructed in wood, with metal panels attached. In addition, the technique of gilding corresponds to what has been identified as a Tibetan gilding technique, different from Sogdian or Chinese methods of mercury gilding. 19 The dimensions of the reliquary excavated at Dulan are approximately 2 times as big as contemporary Chinese reliquaries - the length of the side panels is ca. 25 cm while the height of some fragments is as much as 30 cm. Silk had been attached to the metal, creating an effect of relief. The base fragments with the iron birds were 44.5 cm in length. Such distinctions in gilding process and scale of dimension may be indicative of Tibetan provenance, although possibly craftsmen of other nationalities were working in the Dulan region.

Haarh has studied in detail not only the burial customs but also the mythology of the sPu rgyal dynasty. He has emphasized the ancestor Bya Khri btsan po and the importance of bird characteristics in the literary accounts of the ancestors of Srong btsan sgam po, such as birds' eyes in conjunction with webbed feet, turquoise eyebrows, whiskers and teeth.20 The birds on the Dulan reliquary do not seem to relate to such a description but the interior cavity of the container had a metal base with a rectangular space in the center, into which a bone relic would have been inserted. In comparison to Buddhist reliquaries known in Tang China, the Buddhist ritual function of the container thus seems confirmed, although the decoration neither recalls Buddhist emblems nor the images of Tibetan sacred bird characteristics understood from Tibetan literary accounts. The bird greatly resembles certain designs on textiles excavated from several tombs in Dulan region, which perhaps were copied by local craftsman in metal (plate 7)?21  It is intriguing that the Buddhist reliquary was buried under a boulder just like the live horses, yet the box was burned. It is speculation but perhaps fragrant substances were a burned offering made in the space of the container which was then placed underground, the burial of course gradually extinguishing the fire.

To conclude, I would like to briefly present a drawing on bone which was recovered from the Kexiaotu excavations in Dulan county, also attributed to mid-8th to early 9th century. Plate 8 presents a photograph of a skull previously identified as camel skull but which has now been re-identified as skull of a equid.22 The camel is virtually unknown in Tibetan literature, while the horse as known from the Dunhuang ritual served to guide the deceased. On both skulls may be seen drawings of what Tibetans now call a linga, a ritual offering of anthropomorphic or animal shape in clay or drawn on paper.23 As far as I know, in consideration of the chronology of the tombs, these are the most ancient linga known today, yet they are strikingly similar to the linga of this 17th century manuscript from Lhasa (plate 9), and to linga drawn or printed on paper today.24 The inscriptions are relatively simple: crush the horse demon : rta sri mnan, interspersed with what appear to be mantra.25 The attachment of the linga by a chain at ankles is clear, and the wrists appear bound together, although no cord is visible. This linga so clearly corresponds to models known in later Tibet that it is uncanny to find it excavated in strata of 8th to 9th century. However, the Xining archeologist Tang Hui Sheng confirmed that it was found inside a chamber of a tomb, and considers this to be contemporary with the 9th century strata of this tomb.26 To my knowledge, no additional documents were recovered in the same zone. The ritual which governed usage of this offering is unknown at present, although the striking similarity to later examples would tend to indicate a ritual somewhat analogous to known linga offerings for appeasement of malefic forces. In this case, in terms of sympathetic magic, it would appear that the linga were drawn on horse skulls in order to protect other horses from illness or dire conditions attributed to the sri demons. Not far from this tomb, tsha tsha representing stupa were found at the bottom of the foundations of the peripheral wall which surrounded this tomb complex, which had included a rdo-ring stone stele 300 cm. in height, and two stone lion statues also within the peripheral wall.

To conclude, the excavation of such artefacts in Dulan county provides concrete indications of the syncretic religious practices then observed during burial rituals, which Tibetan literary and historic traditions had preserved. The tombs further document the basic uniformity of the architecture of Tibetan tombs of this period throughout regions under Tibetan sovereignty. The artefacts and tumuli complex revealed by the archeological investigations of Dulan county incite reflection on the antiquity of the Tibetan concepts of sub-divisions of space as well as the antiquity of syncretic practice of Buddhist bone offerings accompanied by animal dismemberment in burials.


[Missing Image]
Plate 1

[Missing Image]
Plate 2

[Missing Image]
Plate 3

[Missing Image]
Plate 4

[Missing Image]
Plate 5

[Missing Image]
Plate 6

[Missing Image]
Plate 7

[Missing Image]
Plate 8

[Missing Image]
Plate 9


Belleza, J.V.

2000 Pre-Buddhist Archeological sites in Northern Tibet : an introductory report on the types of monuments and related literary and oral historical sources (Findings of the Changtang circuit expedition 1999), Kailash, 1-142.

Caffarelli Mortari Vergara , P.

1997 Architectural style in tombs from the period of the Kings, in J.C. Singer and P. Denwood, eds. Tibetan Art Towards a Definition of Style, London: Laurance King, 230-241.

Chayet, A.

1997 Tradition et Archéologie, notes sur les sépultures tibétaines, in H. Krasser, T. Much, E.Steinkellner and H.Tauscher, eds., PIATS Graz, Vienna: Verlag der Österreichischen Akademie der Wissenschaften, 131-139.

Doar, B.

1996 Introduction, China Art and Archeology Digest 1 ( 3), 5-6.

Haarh, E.

1969 The Yarlung Dynasty. Copenhagen: G.E.C. Gad.

Heller, A.

1998 Two inscribed fabrics and their historical context: some observations on esthetics and silk trade in Tibet, 7th to 9th century, in K.Otavsky, ed. Entlang der Seidenstrasse. Frühmittelalterliche Kunst zwischen Persien und China in der Abegg-Stiftung, Riggisberger Berichte V:95-118.

In press Preliminary remarks on the archeological investigations of Dulan: 8th-9th century Tibetan tombs?, in E.Sperling, ed. Proceedings of the International Association for Tibetan Studies, 1998 Bloomington seminar, Indiana University Press.

2003 Archeological Artefacts from the Tibetan Empire in Central Asia, in Orientations, 34/4:55-64.

Huo Wei

1995 Xi zang gu dai mu zang zhi du yan jiu (Tibetan Burial systems), Chengdu: Sichuan University Press.

Kvaerne, P.

1985 Tibet Bon Religion. Leiden: Brill.

Lalou, M.

1952 Rituel Bon-po des funérailles royales, Journal Asiatique, 339-361.

1958 Fiefs, Poisons et Guérisseurs, Journal Asiatique, 157-201.

Otavsky, K.

1998 Zur kunsthistorischen Einordnung der Stoffe, in K.Otavsky, ed. Entlang der Seidenstrasse. Frühmittelalterliche Kunst zwischen Persien und China in der Abegg-Stiftung, Riggisberger Berichte V: 119-214.

Stein, R.A.

1957 Le Linga des danses masquées et la théorie des âmes, Sino-Indian Studies, V, 3-4.

1970 Un document ancien relatif aux rites funéraires des Bon-po tibétains, Journal Asiatique, 155-185.

Xu Xinguo.

1996 (edited and translated by Susan Dewar and Bruce Doar as follows) The Tibetan Cemetaries in Dulan County : their discovery and investigation, China Art and Archeology Digest 1 (3), 7-12.

An investigation of Tubo Sacrificial Burial Practices, CAAD 1 (3), 13-21.

The Buildings for Worship and Sacrifice above Tubo Tombs, CAAD 1(3) ,:23-36

A Silver Casket from a Dulan Sacrificial Horse Trench, CAAD 1 (3), 37-49.

LDe'u Jo sras and lDe'u chos byung, written in 1376 A.D. (1987 reprint, Beijing).


* Research Associate, CNRS Paris, UMR 8047

1 Doar 1996, 6.

2 Heller, in press discusses comparison of architecture and spatial organization of Dulan tombs with tombs in central Tibet and western Tibet ; cf. Caffarelli Mortari Vergara 1997 and Chayet 1997 for detailed analysis of the latter configurations.

3 All dimensions and data on artefacts according to Xu 1996, 24-26.

4 Haarh 1969 : 386-391 discusses the quadrangular gru-bzhi foundation and the cruciform construction dra-mig or re-mig.

5 Haarh 1969: 374 for hypothesis of funerary ritual disposition of utensils, food, attributes

6 Heller in press reviews the question of 2 or 3 levels in tombs according to Tibetan literature; in fact, including the outside "house", this M1 tumulus had 3 levels above ground and no subterranean level.

7 Haarh 1969: 381 cited the Dunhuang chronicle: gyang ro bla 'bubs kyi mgur du bang so brtsigso.

8 lDe'u Jo Sras 1987: 108 de nas bang so phul skyes te, rtsig pa sogs khar byas pas ming mi gung ri sogs kha bya ba yin no, (then when constructing the tomb-offering, after the walls are made as a fortress, it is called by the name mi gung ri sogs kha.) and LDE'U 1987: 251 dur du bang sor gshegs khar (as the burial tomb, the fortress of the deceased) to describe the tomb for 'Bro snyan lde ru, great-grandfather of Srong btsan sgam po, who was buried alive with his two wives. Cf. also Haarh 1969: 334.

9 Macdonald 1971, 254, citing PT 1287, lines 264-274; Chayet 1997, 132 recalls the sBa bzhed prohibition of horse sacrifice during the reign of Khri srong lde btsan.

10 XU 1996, 15.

11 Haarh 1969: 381.

12 XU 1996, 10 and Xu personal communication, March 1998.

13 Stein 1970, 169-170 for the citation of Chinese sources.

14 cf. detailed reconstruction of the texts by Haarh 1969: 368-370.

15 Lalou 1952, 341.

16 Huo Wei 1995: 116.

17 For Haarh's reflexions on gnam sa 'og , see Haarh 1969: 161. Belleza 2000, 41-43 also documents such elevated tomb constructions in northern Tibet. I thank Charles Ramble for kindly sending me this article.

18 Hybrid animals are frequent in contemporary Sogdian textiles and silver objects which circulated in Dunhuang and along trade routes then controlled by Tibetans, see Otavsky 1998. Carter 1999, 31-33 discusses hybrid animals on Tibetan silver attributed to the sPu royal dynasty.

19 Carter 1999, 30.

20 Haarh 1969: 210.

21 Zhao Feng 1999: 114-115, passim. Otavsky 1998 and Heller 1998 discuss the trade routes and provenance from either Sogdiana or Sichuan of these silk textiles, some of which were inscribed with Tibetan words and may have been commissioned for the Tibetan market. See Heller 2003 for silver saddles and birds recovered from another Tibetan tomb.

22 I thank Dr. Gervaise Pignat, Université de Genève, for consulting Pr. Louis Chaix, curator of archeological zoology at the Museum of Natural history of Geneva. In his opinion, these skulls are possibly those of a donkey, a horse or a mule, but without measurements, it cannot be determined more precisely.

23 Cf. Stein 1951.

24 Cf. Karmay 1988: plates 45-46. I thank Anthony Aris for authorization to republish this photograph.

25 The text of the mantra on the outer rim would appear to be as follows: gong sri mar mnan/ rta 'dre dang rta sri mar mnan/.The inner two rims are largely effaced and appear to repeat several syllables of the previous mantra. On the other linga, the outer rim appears to read as follows: las so/ rta la thug (?) thug (?) tsham tshe..drag.um um/ In the small triangles of the next rims, several times rta mnan and rta sri mnan are legible although the entire inscription is again so largely effaced as to preclude a fuller reader.

26 Personal communication from Pr. Tang Hui Sheng, september 1997.

© The Author/s

Actualizado el 24/07/2004

Valid HTML 4.01!
Any Browser!
Creative Commons License
Como lo hacemos?