| articles

by Amy Heller

April 07, 2006

This article was first published in the Tibet Journal, Vol XXIX, No 3; Autumn 2004, and is presented here with the kind permission of the editors.

See The Lhasa gTsug lag khang ("Jokhang"): Further Observations on the Ancient Wood Carvings
by Mary Shepherd Slusser for a commentary on this article. (Link will open in new window)

Please click here to see a Preface to the version of this article which addresses some of the points raised by Mary Slusser in her article.

(click on the small image for full screen image with captions. All photos by author unless otherwise noted)

Fig. 1
In the ninth century inscriptions on the Karchung rdo ring [1] , the foundation of the Lhasa gtsug lag khang, the most revered Lhasa temple, is attributed to the reign of Srong btsan sgam po. The likely period is between 641, when the Chinese princess to whom he was betrothed arrived in Lhasa, and ca. 649-650, when he died. Traditionally, the foundation is attributed to a Nepalese princess, also considered a royal bride of Srong btsan sgam po. Although the historic existence of the Nepalese princess has long been questioned, there is ample evidence to support active exchanges— political, cultural and commercial—between Tibet and Nepal at this time.[2] The Tibetans had encountered the marvels of Buddhist art as an indirect result of their military expansion towards the Himalaya as well as to the Silk Routes and China. Tradition would have us believe that the Nepalese bride founded a temple like those of her homeland. Demiéville however pointed out in 1952 that the Tang annals describe the construction of a residence, Chengyi, to lodge the court.[3] Comparison with the rGya ma residence of Srong btsan sgam po (Fig.1) is important due to the tower core construction of this building which may relate to the original architecture of the central portion of the Lhasa gtsug lag khang as will be discussed hereafter. Despite the reference of the Chinese term to a non-religious structure, still, this might correspond to the apparent core construction of the Lhasa temple because the contemporary model of Nepalese vihara fulfilled both political and religious functions. Although no seventh to eighth century temples are extant today in Nepal, depictions of Nepalese structures in manuscripts and archeological investigations show that such buildings included both chapels and sleeping chambers disposed around a square courtyard, with a series of open assembly halls on the ground floor, and a gallery on an upper floor.[4] This plan was known from India where the fifth to eighth century stone constructions of Ajanta and Ellora are characterized by a series of rooms built round an open courtyard with the room opposite the entryway, which serves as the shrine of the monastery, slightly larger than the other rooms. Adaptations to the Nepalese model rendered in wood rather than stone, were necessitated by the Tibetan climate, notably the stone masonry of the walls, the roof over the open courtyard and wooden pillars for its support. As known today, the most ancient section of the gTsug lag khang seems remarkably faithful to the square form of ancient Nepalese models (see plan, Fig.17 below).
It is beyond the scope of this paper to attempt a precise chronology of the foundation of the gTsug lag khang and its successive phases of reconstruction and embellishment. However, according to several modern historians, many of the activities attributed to the seventh century Chinese bride Wengcheng would in fact be those of the Chinese princess Kimcheng, who arrived in Tibet ca. 710. Her foundation of several temples is historically attested, as is the arrival in Tibet of many artists and religious masters from Nepal, both Newars from the Kathmandu Valley and Indians who reached Tibet via Nepal. Notable examples of this type of migration are recorded in traditions concerning the construction of Samye, ca. 779. Shantaraksita, for example, was summoned from Nepal, and Nepalese artists participated in the initial decoration of this monastery. Although few works of art concretely dated to the seventh and eighth century survive today in Nepal for comparison, the monumental image of Vishnu dated 643 A.D. being a notable exception as well as the basreliefs of the stupa of Chabahil, Tukan-bahal and portions of Sanku sanctuary, numerous small stone caitya in Kathmandu and the corpus of extant Licchavi sculptures allow a sufficient understanding of the tendencies of Nepalese aesthetics and architecture of this period.[5] Nepalese aesthetic elements may be discerned in the door lintels, pillars and capitals in the gTsug lag khang. I propose to examine here a sample of these wooden carvings at different stages of preservation, restoration, and during conservation efforts.


Fig. 2
After crossing two outer courtyards, one arrives at the veritable entrance to the Lhasa temple, on the west facade. The massive wooden doors measure approximately 4 m high. They open onto a corridor approximately 2 m wide, the walls of which are covered with a thick coat of brightly varnished plaster. The very first pillars show traces of successive reconstructions. One might call this a tradition of architectural recycling, as in figure 2, in which one can see the base of a pillar incorporated into the wall which has replaced the pillar as the support of the roof. The existence of the pillar allows us to understand that the walls of this corridor are probably a later stage of construction—initially, it would appear that the doorway had two rows of pillars without walls, supporting a portico roof, as is known from a manuscript illustration dated to 1015.[6]

The sculpture on the base of this pillar is half buried. Stylistically, there is strong Nepalese influence, and the theme itself is a form of Avalokiteshvara especially popular in Licchavi art. Still it cannot be confidently determined whether or not it was part of the initial construction. According to the research of Anne-Marie Blondeau and Yonten Gyatso, the installation of the lateral chapels of this entrance corridor was decorated during the 12th century with the Four Guardian Kings as part of restoration efforts by Gung thang Lama Zhang (1123-1193), so the corridor was certainly in existence at this time.[7]

Fig. 3

Fig. 4

Fig. 5

The principal square of the ground floor of the temple comprises approximately 20 chapels, of which 5 are visibly the oldest. The first floor has the same configuration. Despite noticeable differences among them, to which I will return, these ten chapels do appear older by the sculptures of the lintels, the shapes of the doorways, as well as the construction techniques. On the eastern wall, narrative panels are sculpted as traverse sections of the doorway lintels, while the chapels on the north and south walls have a distinctive structural decoration and geometrically carved lintels. On the northern chapel, the exterior of the doorway has griffons or sardula, animals known from the Indian thrones of late Gupta and Pala-Sena sculptures.[8] On the south, the griffon is replaced by carvings of a lion’s head. There is an intriguing series of dancers and musicians, such as a drummer with monkey head, dwarves eating and dancing—this series of joyous companions at the entrance of temples is found in many Ajanta cave temples (Fig.3) and Lhasa doorway. There has been substantial restoration of these figures, as is demonstrated by this photograph from 1993 (Fig.4), indicating replacement of certain wooden elements. Examination of the base of the doorway pillars reveals that there have been earlier architectural modifications, resulting in the raising of the floor level. On this door, the feet of the sculpture on the pillar base are now hidden from view (Fig.5), while an iron rod has been placed to stabilize another pillar. Despite a slight difference in scale, there is a noticeable similarity to the iconography of the pillars hidden in the walls of the entrance corridor and they could possibly be constructions of the same period.

Fig. 6

The principal chapel of the entire sanctuary is called the Jo khang because it houses the image called Jo bo Rin po che, a statue of the Buddha traditionally attributed to the time of the foundation of the temple. This image is so important that the entire sanctuary is sometimes referred to as the “Jo khang” although this name instead specifies the principal chapel, situated at the midpoint of the eastern wall. Due to the height of the Jo khang, it occupies the space on both ground floor and first floor. We will discuss the Jo khang in detail hereafter. The Jo khang is flanked by chapels whose upper and lower doorway lintels appear visibly older than the other portals, each lintel is divided into narrative scenes. On the first floor, the narrative scenes are limited to the lower register of the lintels, while the upper register is divided into five niches, each with a carved portrait of a single figure. (cf. infra for discussion of the central chapel). Figure 6 indicates both the general appearance of the intriguing narrative panels and also details such as this crowned man, sitting in the middle of a rock cavern. In the same panel, a dancing man holds a sword over his head, next to a Buddha making the gesture of protection. To the right of the Buddha, a woman kneels in adoration. On the door lintels of the chapel to the right of the central chapel, in the upper register, there are scenes of preaching, but the lower register, hidden from view by chains, has even more enigmatic scenes (Fig.7). These include carvings in the lintel representing two dancers, a beautiful woman, followed by a peacock, and, for all appearances, a royal couple in conversation. A useful comparison can be made between these scenes and a small statue from a bas-relief of the Chabahil stupa in Kathmandu (Fig.8), probably to be dated to late seventh to eighth century. In the Lhasa temple lintel, the cylindrical earring, the gently modeled carving and body proportions all suggest a clear relation with Licchavi sculpture.[9]

Fig. 7

Fig. 8

Fig. 9

Several scenes might represent the Buddha engaged in teaching but in other scenes the principal figure bears accoutrements of royalty and not monastic robes. In one, the royal figure appears to be addressing an audience of several different animals, while another scene on the same lintel shows the royal person placing a hare into a cauldron amidst flames, and provoking the flight of the other animals (Fig.9). Narrative scenes on door panels is also a decorative theme known in India, notably episodes from the Ramayana in narrative panels on the Kailasha temple at Ellora dating from the mid-eighth century,[10] However the themes depicted on the reliefs in Lhasa do not derive from the Ramayana despite somewhat similar coronation scenes. One of the literary sources for the Lhasa gtsug lag khang panels may be the Gaṇḍavyuha-sūtra, the 70 chapters of which recount the pilgrimage towards Vairocana as Enlightenment by Sudhana, a young Indian prince. Along the way, episodes similar to the ones found on the panels are narrated, such as the prince receiving instruction at the feet of successive human and divine teachers, as well as his languorous conversations with goddesses in the night or courtesans.[11]


Fig. 10
The choice of the Gaṇḍavyuha to illustrate the Lhasa temple is an element which tends to reinforce the ancient date attributed to these panels. The Gaṇḍavyuha is probably of South Indian origin, but the Avataṃsaka-sūtra cycle which it concludes was probably compiled in Khotan and then translated into Chinese, becoming well known in China as of the seventh century. It was certainly known in Dunhuang during the Tang dynasty. Some episodes are represented in Dunhuang wall paintings,[12] made both before and during the Tibetan occupations (670-690 and 787-866). The translation of the Gaṇḍavyuha-sūtra into Tibetan must have been accomplished during the early diffusion of Buddhism in Tibet. This can be determined because of the quotations of its concluding verses (the Bodhicāryapranidhāna, bZang spyod) as consecration formulae for a sculptural mandala of Vairocana and the Eight Bodhisattva commissioned in 806 A.D. in ’Bis mda’near Jyekundo.[13] Also, a Tibetan translation of the Gaṇḍavyuha is listed in the Tibetan royal library catalogue of ca. 822.[14] The popularity of this text in Nepal cannot be documented earlier than the late 11th or early 12th century, the approximate date of a manuscript found by Professor Tucci in Tibet and now dispersed in private and public collections. [15] In conjunction with these Nepalese manuscript illuminations, the narrative panels of the Gaṇḍavyuha sculpted in the early ninth century at Borobudur, Indonesia, facilitate tentative identification of some of the episodes represented in Lhasa.[16] Even allowing for the differences in media and techniques (painting vs. sculpture), it is striking to compare the body proportions and stance of the figures from this illustration of an episode from the Gandavyuha manuscript now in the Los Angeles County Museum with this panel from the Lhasa gtsug lag khang (Fig.10)

As only a relatively limited number of narrative panels are extant today, the hypothesis of this identification must remain somewhat tentative. Also, it is not being claimed that the Gaṇḍavyuha was the exclusive textual inspiration, since some panels are apparently related to the Amitayur-dhyāna-sūtra (Fig.9). One seems to narrate an episode from the story of King Bimbisara already represented during the seventh-eighth century in Dunhuang with this episode of immolation of a hare[17]. This sutra was also in the catalogue of the Tibetan royal library; it was certainly translated very early and known in Tibet. While the extent to which the Jataka tales were known in ninth century Tibet remains open to question, there is virtual certainty that the Gaṇḍavyuha was popular in Tibet at this time, due to the cult dedicated to Vairocana by the Tibetan sovereigns, hence the great likelihood of its representation in the principal Lhasa temple.[18]

Fig. 11

The study of the panels for their iconography has led to an unexpected discovery regarding their structure. Most of the lintels are composed of a single, solid piece of wood, in all probability a single tree trunk, now approximately 25-30 cm thick. However certain panels have an interlocking edge, which allows insertion of the panel in between adjacent panels on both sides. In other words, some lintels are comprised of both permanent, fixed sections as well as non-permanent fixed elements, which could be moved to other locations, substituted, replaced or renovated over the course of time. For example, in this panel of the eastern wall chapel on the first floor, some details may appear to be characteristic of later iconography, such as the figure who might indeed be a mahasiddha (Fig.11). It seems doubtful that the liturgy of the mahasiddhas became known during the early diffusion of Buddhism in Tibet. In consideration of the Tibetan tendency to copy earlier styles, it is possible that this section of the lintel was a later addition, perhaps carved during restorations known to have occurred during the 11th century.[19]

Before discussing the Jo khang chapel, in addition to questions of later substitutions or replacements, the question of successive cross-influences is pertinent. To what extent did the fame of the Lhasa temple spread during the Tibetan Empire; did it serve as a model? Extant examples in Tibet are few. Perhaps Samye, Tradruk and Keru preserve the outer appearance of the vihara square, but the interior architecture has undergone substantial restorations over the centuries. As a possible emulation of Lhasa, these pillar bases come from a monumental tomb near Chengdu, Sichuan, dated ca. 925-950; in subject matter and body proportion, they are very similar to those of the Lhasa temple,[20] (Fig.12) originally inspired in turn by Nepalese Ganas (Fig.13). This is not completely surprising because in addition to close relations with Nepal, contacts between Buddhists in Tibet and Sichuan were ongoing during the eighth and ninth century,[21] and involved the exchange of texts and teachers. Commercial and political relations linking Lhasa with the Nanzhao kingdom in Yunnan during the Tibetan empire have also been amply documented.[22] Although very few examples have yet been identified, the question of the spread of Tibetan aesthetic influence to Sichuan as well as Yunnan may be raised.[23] Archeological investigations now in progress in Sichuan and Yunnan will hopefully provide more documentation on this subject.

Fig. 12

Fig. 13

Fig. 14a
To conclude, it is important to discuss the ancient carvings at the entrance to the Jo Khang, the chapel which houses the Jo bo statue, the central chapel whose contents are deemed so quintessential that the entire sanctuary is sometimes refered to by the name of this chapel. Bearing in mind considerations of cross-influences and successive restorations, it is to be recalled that Béguin et Mortari Vergara de Caffarelli had already evoked comparison to Ajanta in previous studies, specifically to the architecture of the doorway to cave 2 (Fig.14a). [24] While this comparison may also hold true for doorways on the north and south wings, the central chapel is dissimilar in some respects. The height of the two principal lateral pillars with carved standing figures is just 3 m, but the full pillar height reaching to the floor of the first storey is approximately 10 m, which corresponds to the ceiling height inside the Jo bo chapel. The carved section of the pillar is surmounted by several tiers of curtains suspended across the doorway; a series of shelves for small statues or books occupies the width of the doorway up to the ceiling. In support of the agreement between the CNRS and the Lhasa Academy of Social Sciences, the monks gave a special authorization to allow me to view behind the curtains bringing a very high ladder. The wall behind the curtains reveals three horizontal rows of small friezes of amorous couples in embrace (fig 15a). These are unlike any other friezes in the temple and are exceptionally well conserved beneath layers of paint. To a certain extent these figures correspond to the Gandarva-Apsara couples frequent in India, [25] such as those painted on friezes placed above the doorway at entrance to Ajanta cave 17 (Fig.15b). Although the iconographic scheme above the Jo khang recalls Indian antecedents, the carving of these is manifestly Licchavi in body proportions, garments and jewelry replete with scarf at the neck and cabochon studded crowns. In a slightly larger scale, this bracket adjacent to the Jo bo chapel, illustrates the same aesthetic paradigm of the carved Gandarva, albeit alone; it may indeed be probably contemporary (Fig.16). Indeed, aesthetically and structurally the doorways and brackets of the eastern wall would seem to represent a unified concept which is distinct from the decorative scheme and architectural structure represented by the doorways of the north and south walls. The ceiling height of the Jo khang chapel on the eastern wall extending to the upper floor, is approximately 10-12 meters, with the four lateral chapels and the central sanctuary forming a sort of core tower, around which additional chapels seem to have been constructed later. The construction of a core tower with subsidiary chambers recalls the architectural models of rGya ma (Fig.1) and Yum bu bla mkhar, traditionally believed to be the first royal residence. [26] As a working hypothesis, the following outline is proposed for the initial phases of construction (Fig.18):
ca. 640-650 residence
ca. 710-750 reconstruction of core tower as central chapel and lateral chapels post 780-835 adjacent sleeping quarters progressively linked to core tower
ca. 1050 reconstruction of north and south walls following Indian portal style for central chapels of each wall on both ground floor and first storey
ca. 1075 construction of the Zhal ras lha khang, according to scheme by Roberto Vitali [27]
ca. 1160 embellishment of the entrance corridor by Gung thang bla ma

Fig. 15a

Fig. 15b

Fig. 16

Fig. 17

Fig. 18

I would like to acknowledge financial support by two anonymous private donors and the CNRS URA 1229 for two trips to Lhasa in 1995 and 1996. In Lhasa I was helped by Phuntsog Tshering, President of Lhasa Academy and Lama Phurbu of the gTsug lag khang. For fruitful criticism, I particularly thank Valrae Reynolds, Lionel Fournier, and Ernst Steinkellner. Grateful acknowledgement to Rob Linrothe for editing, and to Ian Alsop and Corneille Jest for critical reading.


1. Richardson 1985, 72-81. Richardson 1977 provides the first detailed discussion of the entire Lhasa gtsug lag khang, while Shakabpa 1982 gives summary of the history of the temple (7th to 20th century) and detailed inventory of contents of all chapels.

2. For bibliography from 1950 to 1996 of historians addressing this issue, and absence of all mention of her prior to 11-12th century Tibetan historical works such as the Maṇi bka’’ bum, cf. Heller 1997. H.E. Richardson has discussed this most recently, cf. Richardson 1997. Slusser 1982, 33 investigated the Nepalese historical sources and found no corroboration of the matrimonial alliance of Srong btsan sgam po and a Nepalese princess.

3. Demieville 1952, 201.

4. Locke 1985, 4. Slusser 1982, fig.244, a vihara depicted in a Nepali manuscript dated A.D. 1015 shows the two-storey construction.

5. cf. Slusser 1982, plates 254-288 are photographs of Licchavi caitya of the Kathmandu Valley. Cf. PAL 1974, plates 8,9,12,14 Passim and Bangdel 1989.

6. Slusser 1982,. fig.244, vol. 2.

7. personal communication from A. M. Blondeau. Cf. Blondeau and Gyatso 1997, 47-49.

8. Auboyer 1949.

9. for chronology of Licchavi art, cf. Pal 1974 and Bangdel 1989. For a stone sculpture with identical cylindrical earring, cf. Pal 1974, fig.227, attributed to 10-11th century.

10. Gail 1985.

11. cf. Richardson 1990, 271-274; Cleary 1989, and Fontein 1967.

12. Wu 1992.

13. Heller 1994, 74-79; Heller 1994, 335-349

14. Heinkellner 1995, 14-19; Lalou 1953.

15. Huntington and Huntington 1990, pages 251-263, plates 89a-d.

16. Fontein 1967; the complete series of all Borobudur sculpted panels identified by Frederic and Nau 1994, cf. in particular p.259, fig. 11.68 for the seated couple of Suchandragrihapati (?) 48th master, as well as the 17th master, Samantanetra, reclining amidst a group of animals, p.307, pl 11 b 29;).

17. Whitfield and Otsuka 1996: 313-316. To summarize the episode: King Bimbisara longed for an heir and learned that a hermit would be reborn as his son. Having begged the hermit to die early, he refused, and then changed into a white hare that was killed by the king. The soul of the hermit was then indeed reborn as the crown prince.

18. We are indebted to Yoshiro Imaeda for confirming this information on the Gaṇḍavyuha, cf. Imaeda 1981., cf. also Richardson 1990.

19. Vitali 1990, 78-83.

20. Archeological Team 1991.

21. Broughton 1983.

22. Backus 1981, Lutz 1991.

23. The only extant sculptures of the Nanzhao royalty wear robes virtually identical to those of statues of Vairocana and the eight Bodhisattva carved near Jyekundo in 806 A.D cf, Heller 1994, Heller 1997.

24. Caffarrelli and Beguin 1987, 258-260..

25. I thank Lionel Fournier for this iconographic identification.

26. To my knowledge, Chab spel 1984, 14, first discussed the idea of the core of the temple as mkhar, a tower-fortress. Mi nyag Chos kyi rgyal mtshan considers this hypothesis of mkhar core structure for the Jokhang as architecturally cogent. As part of the restoration of the Potala, he discovered remains of a series of mkhar which he attributes to the time of Srong btsan sgam po, discussed in his presentation at the 1998 IATS seminar, Bloomington, cf. forthcoming article Minyag.

27. Vitali 1990, discusses 76-83 a construction of a new chapel, the Zhal ras khang, in 11th century. The ancient doorway to the chapel was replaced during 1990’s restoration of the Lhasa gtsug lag khang and had not been photographed beforehand to my knowledge. Due to these factors, it has not been possible to include examination of its lintels in this study.

REFERENCES | articles