Ancestors, Politicians, and Patrons: Portraits of the Dead in Ninth and Tenth Century Dunhuang Mogao Caves

MAPH THESIS AWARDS 2013 By Madeleine Boucher


I. Introduction: the dead and the living together in the 9th and 10th c. Buddhist cave temple

The almost five hundred Mogao Buddhist cave temples (莫高窟) built into the cliff face near Dunhuang (敦煌) constitute an unparalleled panorama of Chinese Buddhist art, representing a span of nearly one thousand years, from the early 5th century CE to the 14th century CE during the Yuan Dynasty. From the time of the earliest caves cut at the Mogao cave temple site, located in northwest China just outside the former border city of Dunhuang, powerful local families, clergy, and laypeople of more modest social standing alike sponsored cave building as a form of religious devotion, as well as to achieve more worldly ends by publically displaying that devotion.

Following a long Buddhist tradition of visually recording donations, these donors had their images and images of their relatives, both living and dead, painted onto the cave walls. These donor-worshipper images[1] play many roles simultaneously: they mark the act of donation, recording the donor’s act of devotion; they act as simulacra-worshippers standing in for their real-life antecedents by remaining inside the cave to worship its sculptures and painted icons in perpetuity; and some of them visually represent the donation’s beneficiary, who was often deceased when a family member made the donation on their behalf. Consequently these donor-worshipper images are not just portraits—they are also indexes of donation practices that tied the living and dead together as worshippers and family members, as well as symbols of the continued presence of the people they represent (by now, all long deceased) inside the cave space.

While including representations of the dead among donor-worshipper images was a long-established practice by the ninth century, these representations later developed into several distinct forms of portraiture practice that engaged with the viewer in a way that was unprecedented in the Mogao caves. In earlier Mogao caves, representations of the deceased appear in family groups of donor-worshipper images side by side with their living relatives, distinguished only by the inscriptions that mark them as dead. Precedents for the practice of depicting the dead can be found far and wide, from roughly rendered figures carved into sixth century Buddhist stone steles from the Chinese heartland to the famous relief sculptures of an imperial procession from the Binyang Central grotto at Longmen, which depicts the late emperor, the beneficiary of his son the current emperor’s donation to construct the grotto.[2] In the waning years of Dunhuang’s Tibetan period (768-850 CE),[3] the patrons and artisans who realized the caves began to include new types of donor-worshipper images that literally and figuratively elevated the dead to new heights of prominence in the cave space.

First, in a handful of family caves, patrons included portraits of their deceased parents or other departed relatives above the east wall doorway, literalizing the dedication to deceased ancestors that had long been a major motivation for Buddhist commissions.[4] Second, towering, larger-than-life portraits of deceased political leaders and their living descendants stand in processions lining the corridors of most of these same caves. The viewer’s encounter with these corridor portraits marks the transition into and out of the interior space of the cave, a transition that symbolizes the ties between the sacred space and worldly power relations. At the same time, just as in older caves, there also appear donors who were deceased at the time the cave was built, worshipping alongside their living counterparts in standard donor-worshipper images[5]. While not every cave includes every type of portrait, ancestor portraits, monumental corridor portraits, and standard donor-worshipper images were incorporated into a standard repertoire of cave building conventions during the Zhang and Cao Guiyijun歸義軍 “Return to Righteousness Army” Periods (850 – 914 CE and 914-1036 CE respectively).

Examined in their native context in the cave space, these portraits are part of a broader sea change in the way that cave temples represented patronage, family relationships, social hierarchies, and the relationship between the living and the dead. This change might be described best as an intensified presence and a more literal representation of the worldly relationships that were always inherent in the practice of Buddhist donation. Because constructing cave temples generated karmic merit for the builders that would also benefit their deceased and living family members, the act of building caves reinforced family ties between the living and the dead. Like merit, which aligned Buddhist beliefs with traditional displays of filial piety, commemorating one’s family across the boundary of death might be understood as a reflection of Buddhist beliefs about rebirth, beliefs that mobilized the traditional Chinese conceptions of a multi-generational family unit extending into the afterlife. Such an emphasis on lineage, continuity, and familial unity also foregrounds one longstanding purpose of Buddhist caves: to commemorate the dead while testifying to the power of the living families who built them.

While each type of portrait described above is different in its form and function, all were conceived as integral elements in a cave space that became increasingly large, lavishly decorated, and filled with markers of domestic luxury; in combination, these elements created an encyclopedic Buddhist paradise on a small scale—a virtual facsimile of a microcosmic Buddhist world. Thus, architectural space is the unifying structure that both constructs the roles of the figures depicted within it and ties these images together into a complete experience for the viewer—an experience, that is, of the entire architectural space of the cave. Embedded in the very structure of these portraits of donor-worshippers is their relationship with the worshipping viewer, which they create by virtue of their formal characteristics, their composition, and their architectural placement within the cave. Without taking on the seemingly impossible task of reconstructing the perspective of a ninth or tenth century viewer, it is still possible to piece together textual evidence and the physical evidence of the cave’s layout to imagine how a visitor would have experienced the space. After all, the experience of the worshipper is already present in the wall paintings themselves, since these donor-worshipper images bear witness to the cave interiors and remain active participants in worship.

By examining the formal qualities of these images in relation to their architectural context and by analyzing the compositional structure of these portraits of laypeople with reference to the position of the viewer, I will argue that the portraits formally interpellate a worshipping viewer. Rather than simply acknowledging and including the position of the worshipping viewer within their design (interpolating the viewer, in other words), the portraits also engage with the viewer in a dynamic relationship and bodily situate him or her within the cave space more broadly. Put simply, the painted figures on the walls guide the viewer’s attention to objects of worship and locate this viewer within the paradisiacal realm of the Buddhist cave temple.

The harmonious coexistence of the living and the dead in these wall paintings marks the cave space as one of suspended time and space apart from, yet closely engaged with, the outside world. In this sense, these images of real people, both living and dead, bridge the gap between viewer and the cave space—they are familiar and recognizable real-world entities, and their power to address and engage with the viewer transitively pulls the viewer out of the mundane and into the paradisiacal world that they inhabit. This connection is less intellectual or psychological than it is visceral—it is constructed through the physical relationships between bodies both living and painted, between types of spaces and movement that are both spiritual and material at once. Similarly, we can understand the cave space as one where sacred power inheres, in its architecture, its paintings, and its sculptures all at once. The cumulative effect is an ‘actualized’ totality of the paradisiacal realm— the cave space as a whole, in other words, does not merely evoke the concept of a Buddhist paradise, but rather, the cave itself actually manifests the material qualities of a paradise of different spatial and temporal dimensions.[6] Together with the innovative architecture and wall paintings of the ninth and tenth centuries, ancestor portraits, monumental corridor portraits, and standard donor-worshipper images link the viewer to the cave space, enveloping them in the paradisiacal realm of the cave and, in the process, making make them truly present in that other world.


II. Innovations of the ninth and tenth centuries at Dunhuang

During the ninth century, stylistic changes in cave design at Dunhuang accompanied an increase of the size, quantity and diversity of portraits of laypeople. In some caves, for example, donor-worshipper images occupy nearly a third of the wall space, a dramatic increase from that of earlier caves. At the same time, other architectural and stylistic changes in cave building developed, such as a new shape of the Buddha niche, a new mode of partitioning the wall murals into sutra paintings, and a wealth of intricately planned and highly detailed polychrome decoration that epitomizes horror vacui[7].  Along with new modes of depicting laypeople, these stylistic innovations would seem to indicate a shift in the conception of the cave space. As mentioned earlier, this shift could be described as an intensification of the social, familial and commemorative functions that cave temples had already fulfilled for centuries. The fact that ancestor portraits first appeared during this same period seems to be no coincidence.

A number of these stylistic innovations appeared together for the first time in cave 231 (dated to 839 CE), the cave that contains the earliest known ancestor portraits of this kind at Dunhuang. Winston Kyan summarizes the style that cave 231 brought to full fruition:

…the brilliant mineral pigments and intricate Buddhist iconography dissolve the rock surface of the cave, turning it into a vast brocade-lined tent that hovers between the material and the immaterial. But on closer inspection, these textile-like patterns emerge as a carefully planned program of spatial, decorative, figurative, and epigraphic references to the domestic realm that are transposed onto the shell of a Buddhist cave.[8]

In this architectural plan, which would dominate Dunhuang cave-building for the next two centuries, the central Buddha niche in the west wall or Buddha altar in the center of the cave are shaped like a domestic couch-bed (chuang 床) surrounded by painted screen panels (fig. 1). fig 1The rectangular divisions that partition the sutra paintings on the walls of the cave simulate the screen panels that would line a domestic interior (fig.2). In addition, the size of the average newly-cut cave increased substantially, with many measuring ten meters across and ten meters high on each side.  The vast majority of ninth and tenth century caves also follow a standard architectural plan. This plan consists of an “inverted-dipper-shaped” (fudou 覆斗, like a four-sided pyramid with a flattened top cap) ceiling and a roughly square main chamber with one central space of focus where sculptures of the main icons were placed—either a rectangular niche set into the center of the west wall, or a rectangular raised platform freestanding in the center of the chamber. Both types of central spaces for the Buddha or bodhisattva sculptures share the same chuang-like shape with a horseshoe-shaped raised platform for the statues to rest on.

Scholars have often noted the heightened emphasis on rich material decoration—paintings in vibrant blue, green, red, and gold pigments that seemingly transform every available inch of wall and ceiling into patterns or pictures of lavish detail. As Neil Schmid has argued, during the ninth and tenth centuries, cave architecture translated the trappings of private interiors into encyclopedic schemas of the Buddhist cosmos. Schmid views this transition as a conceptual transformation of the cave space into a “kinesthetically and experientially accurate” ritual setting, one that adds a layer of symbolic reference that Schmid insists is related to sutra lectures and the ritual performances that would have occurred in the cave space.[9]  Similarly, in her essay on the multi-panel wall paintings of late eighth and early ninth century caves at Dunhuang, Ping Foong links the screen-format wall paintings and chuang-shaped niche to Tang dynasty tomb architecture of the preceding centuries.[10] She argues that although the domestic elements of painted cave spaces can be traced to direct precedents in palatial and tomb architecture, cave temples employ these visual tropes in a fundamentally different way: representations of domestic furniture in cave wall paintings simulate a familiar setting and thereby situate the viewer in the cave space on a physical level. These representations are therefore less a “secularization” of cave decorative programs than a “sanctification”, or reification.[11] They transport the practitioner, visually and somatically, to the paradisiacal world within the cave. In contrast to Schmid, Ping Foong suggests that the cave space has the power not just to evoke another world but even to manifest that other world as a material and tangible reality, one in which it also bodily situates the viewer. Images of laypeople within the cave space likewise serve to facilitate the manifestation of this tangible reality by suggesting that, just as they do, the viewer has a place in this sacred space. They span the gap, in other words, between the embodied reality of this paradisiacal world and the material world of the viewer that overlaps with it.

The Zhang Guiyijun period saw a frenzy of cave-building activity at Dunhuang, second only to the Sui Dynasty in terms of caves built per year. The subsequent Cao Guiyijun period was the most active period of cave renovation in the history of the Mogao site, perhaps because the cliff face became saturated, thus encouraging patrons to appropriate and remake older caves rather than cut new ones.[12] During this period, the majority of caves were constructed as “family caves”—that is, caves primarily sponsored and owned by a single family, according to Ma De’s definition. The ruling Zhang and Cao clans were closely tied by blood and marriage to other prominent Dunhuang families; these families, especially the Zhang and Cao clans, were the driving force of cave building and renovation. These are the patrons (kuzhu 窟主, cave owners/cave patrons) of many caves, including those that the present study will examine—caves 231, 138 and 156 in particular. Caves were family assets, and building or renovating caves was an expedient method of displaying family power and status as well as Buddhist devotion.

Though the motivations of patrons will not be my primary focus, it is important to bear in mind the central importance of the act of donating to garner merit in Buddhist belief in order to understand why building caves intimately connected family, Buddhist faith, and the commemoration of dead ancestors. Paying to create material objects, such as commissioning sculptures or building structures, or sponsoring other acts of charity, such as contributing funds or goods to support the clergy, generates merit for the donor and his or her named beneficiaries. Merit acts as the currency of a cosmic moral order that rewards good deeds in the present and, after death, rewards believers with rebirth in paradise. In a transaction recorded in accompanying inscriptions, not unlike a contractual exchange, donors could transfer the merit their donation generates to another—most often, to deceased parents and ancestors. As John Kieschnick describes, the complex motivations behind such Buddhist donations were worldly as well as spiritual; not only did the donors and their ancestors accrue karmic merit for a better life or better rebirth, they also fulfilled philanthropic and filial obligations through their donations: “the gift […] in honor of a deceased parent was an important public expression of filial piety and local philanthropy, regardless of one’s personal belief or disbelief in the Buddhist doctrine of merit.”[13] In this sense, the Buddhist doctrine of merit held that material things—Buddhist objects, images, and architecture—had real efficacy to benefit both the living and the dead. Material objects and their production also allowed donors to commemorate the dead in a Buddhist manner entirely compatible with the traditional demands of filial piety, which called for public displays of veneration of one’s parents and ancestors.

The merit record of Yin Jiazheng 陰嘉政, which memorializes his building of cave 231, demonstrates how deeply both the economy of merit and filial piety suffused patronage at Dunhuang. Indeed, the term “merit record” gongdeji 功德記 and the term “merit cave” gongdeku 功德窟 (describing the caves built to generate merit, which were dedicated to a named beneficiary) imply that performing the act of donation was just one part of the equation: memorializing the merit-generating act and rendering it in a material form that would preserve it for others to see was also crucial for this act’s efficacy. Winston Kyan and Sha Wutian 沙武田, among others, have closely analyzed this text (preserved in a manuscript transcription of the original merit record stele that originally stood outside cave 231[14]), and while this study will not seek to replicate their work, a few parts of the text provide important context. The portion of the text that describes the construction of cave 231 reads as follows:


Accordingly, he [Yin Jiazheng] hired good workers, and recruited skilled artisans, and on the second level [of the cliff face], he built a cave. A plaque outside this cave declares it repays the kindness of the lord and the family.[16]


This standard concept of “repaying the kindness ” 報恩 of parents and ruling authorities[17] through Buddhist donation ideally embodies the intersection between Buddhist merit practices and performing the duties of filial piety, which is the subject of Kyan’s in-depth study.[18] By building cave 231, Yin Jiazheng accrued a significant amount of karmic merit, which his merit record officially transfers back to his deceased parents, a filial deed as well as a standard Buddhist dedication to ensure that the merit would benefit the donor’s deceased loved ones. While these sorts of textual dedications had a long history of memorializing and effectively sealing contracts of Buddhist donation over the preceding centuries, the ancestor portraits of cave 231 were perhaps the first to make the dedication directly visible. Ancestor portraits thus kept the donation to “repay the kindness” of parents perpetually present in the cave space in an innovative visual manner.


III. Part 1— Ancestor Portraits

By literalizing the symbolic presence of the dead in the cave space, ancestor portraits in these ninth and tenth century caves became material markers of centuries-old ideas of merit generation and filial piety through Buddhist donation. Not only do ancestor portraits physically place the traditional beneficiaries of the donation, the cave patron’s deceased parents and his other forbearers in the cave space, they also frame these ancestors as objects of worship through spatial placement and pictorial parallels to the Buddha niche. Highly individualized, these figures are also worshipping subjects whose naturalistic representation invokes their physical presence. In this sense, ancestor portraits concretely literalize bonds between family power, worldly authority, and cave patronage that had always been present in the history of cave building, making them an important part of the viewing experience.

There are approximately a dozen caves at Dunhuang in which representations of the donor’s dead parents appear over the doorway on the East wall.[19] Of these caves, cave 231 is the earliest in which this type of image appears—it can be reliably dated to 839 CE based on the date of the merit record of cave patron Yin Jiazheng. Later caves emulate the format of the ancestor portrait of cave 231 precisely, which might testify to the importance of this cave as a model for later cave building. Though cave 231 might therefore be considered the progenitor of the ancestor portrait in the cave space, it seems unlikely that the ancestor portrait was an entirely new form at the time of its introduction; rather, they were probably adapted from funerary portraits, as my subsequent discussion of the Stein collection paintings will demonstrate.

In a floral patterned rectangular frame above the doorway of the east wall of cave 231, a richly dressed female figure at left and a male figure in formal robes at right kneel on one knee around a decorated rectangular pillar at center (fig. 3). fig 3The figures are turned in three-quarters view facing the central pillar, and both clasp intricately decorated gold-leaf incense burners. These figures are the deceased parents of Yin Jiazheng. Their postures mirror one another as they sit on square platforms with carpet-like patterning, seeming to rest on a plane raised above the surrounding attendant figures flanking the two parents. The attendants wear thickly draped layered robes and hold bowls with what appears to be offerings of fruit or flowers (fig. 4). What is most immediately striking about the portrait is the amount of detail and individual attention invested in the figures of the parents, especially in the mother of Yin Jiazheng, and even in the figures of the attendants (fig. 5). Such attention to naturalistic volume and the shapes of the bodies of the figures—particularly evident in their carefully rendered faces— is unprecedented in depictions of laypeople outside of narrative paintings in the Dunhuang caves. It is only in ancestor portraits that these details seem to give exacting information about the physical appearance of the subject of the portrait, rather than merely adorning the image of the donor.fig 4

The pillar structure standing at the center of the composition appears to be an architectural element, rendered with three-dimensional volume, but also contains the dedicatory inscription for the two figures. In this sense, it serves a dual purpose as both a physical pillar with an inscribed central tablet occupying the compositional space and also as an inscription cartouche. Though they are not explicitly named in the central inscription on this pillar between the two kneeling figures, comparing the titles and lineage of the two figures with the patron’s merit record identifies the deceased parents as Yin Bolun 陰伯倫 and his wife of the Suo 索 family. The inscription reads:




The deceased father, Danzhou Changsong District Officer of the Left of the Metropolitan Guard [and]

The kind deceased mother descendent of a Dunhuang official of the Suo family worship/donate with the same heart.


In his analysis of this ancestor portrait, Winston Kyan notes the unusual dedicatory phrase at the end of the inscription on the central tablet, “同心供養” (worship or donate literally “with the same heart”, figuratively meaning “together”), as opposed to the “一心供養” (worship or donate wholeheartedly) standard in inscriptions of individual figures. The substitution of tongxin 同心 for yixin 一心 is an unusual variation on the standard donor dedication phrase, normally a set formula, that declares the unity of the couple. The visual symmetry of the couple, which is paired with an unusually egalitarian depiction of male and female occupying the same register of space, serves to reiterate this unity.[21] A similar inscription can be found on the tablet in the center of the ancestor portrait in cave 138, which depicts two women and one man kneeling on raised platforms surrounding a central tablet with a large array of attendants surrounding them (fig. 6). The inscription states that together, they “worship/donate with a wise and benevolent nature” zhihuixing gongyang 智惠性供養, again emphasizing their status as a unit of worshippers.fig 6

Scholars seem to agree that the inscribed pillar at center represents an ancestral tablet based on its striking similarity, with its decorative cap and wide base, to the tablets that represent the deceased in memorial rites. Identically shaped tablets also appear at the center of ancestor images in caves 9, 12, 20, 138 and 144. The consistency of its form indicates that the pillar at center represents a real-life object—the shape of the tablet is consistent in each portrait, with the same flared decorative cap, tassels and a wider, carved base. The fact that in funerary practices an ancestor tablet typically would stand in lieu of a portrait of the deceased makes the tablet’s presence among these portraits of the patron’s ancestors all the more enigmatic.[22] The result is a doubling, a representation of representations: the figures are highly mimetic depictions and therefore invoke the presence of the dead through their physical appearance, while the tablet symbolically represents the dead, using writing to memorialize them. The former seems to be a substitute for the parents’ presence, the latter a symbol of their absence.


III. Part 2— Ancestor portraits as a genre and in the cave space

While a number of scholars have addressed select ancestor portraits in isolation or noted their similarity to one another, to my knowledge, no one study has as of yet thoroughly analyzed these portraits together as a genre. Understanding them as a genre is crucial not only because ancestor portraits are so different in formal composition and function from the donor-worshipper images with which they are sometimes confused, but also because viewing them thus brings together many seemingly independent developments in cave building that scholars have already noted. Winston Kyan’s 2006 dissertation and 2010 article, “Family Space”, adapted from that dissertation’s final chapter, masterfully analyze the portrait of the patron’s parents in cave 231 in relation to the cave’s stylistic and architectural innovations. Kyan argues that the portrait illustrates the fusion of family commemoration and religious donation. These ancestor portraits and other architectural innovations (the couch-like niche and screen format wall paintings so like a domestic interior) form a visual vocabulary singularly suited to representing family identity. Kyan concludes that by transforming the cave into a “theater of domestic luxury,” patrons fully realized the function of the Buddhist cave as a family space.”[23] His study lays the groundwork for an expanded reading of the viewer’s experience within this new cave space, which is less Kyan’s concern than the familial and filial relationships represented. I intend to build on his work to better understand the unique qualities of ancestor portraits as a genre and the significance of portraits of the dead in the cave space as a whole.

Dunhuang Academy scholar Sha Wutian has also published studies on a number of individual ancestor portraits, including one on the iconographic qualities of the same portrait in cave 231 in collaboration with Bai Tianyou百天佑. The authors argue that the ancestor portrait in cave 231 has a broader political significance because it proffers a compromise between traditional Tang culture and Tibetan rule in Dunhuang.[24] Sha Wutian, Zhang Jingfeng 張景峰 and Gu Shuyan顾淑彦have also analyzed the ancestor portrait in cave 138  (dubbed the “wise and benevolent nature,” zhihuixing 智惠性portrait after part of its unusual inscription) through a historical lens to identify the patrons and trace the construction of the cave.[25] Nevertheless, while many scholars have remarked on the uniqueness of the portrait of the patron’s parents in cave 231, writings on donor-worshipper images have rarely distinguished between different categories of images of laypeople, often opting instead to treat “donor images” as a monolithic category with a direct correspondence to the social and political conditions of patronage.[26] One of my aims in this study will be to disambiguate the various functions of these images, beginning with ancestor portraits, which, unlike standard donor-worshipper images, construct an iconographic relationship with a worshipping viewer. My broader goal will be to articulate more fully the breadth of roles that such representations of non-religious figures can play by paying particular attention to their ability to bridge the gap between the viewer and the Buddhist cosmos that the cave space manifests.

The consistency of the ancestor portrait format within a variety of different caves implies that the form of these portraits is inseparable from the function they serve in the cave space—their significance universally depends upon their placement in relation to the Buddha niche, the niche-like composition of the portrait space, the individualized portrayal of the deceased, and the tablets inscribed with their names at the center of the composition. Tracing the roots of these fundamental characteristics, Kyan emphasizes the functional similarity between these ancestor portraits on the east wall of caves and the miaozhenxiang邈真像 “portrayed real visages” portraits, which were customarily painted shortly after a person’s death (though some may have sat for theirs during their lifetimes). Miaozhenxiang were installed above temple and image hall doorways to commemorate the dead, a space strikingly similar to that in which ancestor portraits were placed in Dunhuang caves. The parallel seems to confirm the funerary function of ancestor portraits—Zheng Binglin even suggests that based on depictions of dead ancestors in hanging and scroll paintings from the library cave, donor-worshipper images in wall paintings may have broadly served as miaozhenxiang that commemorated the dead[27].

Though these ninth and tenth-century devotional paintings found in Dunhuang Mogao Cave 17, now in the collection of the British Museum, may not be the same genre as the miaozhenxiang mentioned in textual records, many include highly individualized portraits of deceased donors that seem to serve the overtly funerary purposes that Kyan and Zheng describe. One example is Stein painting 14, dated to 910 CE, a copiously inscribed depiction of the bodhisattva Guanyin觀音 (skt. Avalokitesvara) that is remarkable for the two large and detailed figures standing at the bodhisattva’s side (fig. 7). fig 7Inscriptions declare that the painting is dedicated to the memory of the two individuals depicted, giving the devotional image an overtly commemorative and funerary function. The woman represented in rich layers of robes, with a delicate face and bluish shaved head, is the deceased nun Yanhui 嚴會; the other figure of a young layman with a boyish hairdo is the donor’s deceased brother, the probationary chamberlain Zhang Youcheng 張有成[28]. Fascinatingly, the inscription twice refers to the portraits as maozhen貌真, a phrase that in this context was used interchangeably with miaozhen邈真. This undeniably links the portraits to those to which Kyan and Zheng refer, and it may be that the standing portraits in Stein painting 14 are in fact concrete examples of images that doubled as funerary portraits and as Buddhist donor-worshipper images. They represent the dead accurately and individually to preserve their memory, and also to eulogize them as the beneficiaries of Buddhist merit, as Zheng suggests.

In other Stein collection paintings, many other donors both living and deceased can be found in kneeling postures strikingly similar to the ancestor portrait format, proving that the format existed in different media, even outside of the cave space. In ninth and tenth century paintings, it seems that the vast majority of donor-worshippers in the paintings kneel on one knee in three-quarters view at the bottom of the image, with dedicatory inscriptions written in cartouches centered between the figures. Only one such painting has been roughly dated to the same time period as cave 231’s completion in 839 CE, first half of ninth century at latest. Many others—such as Stein paintings numbers 1, 2, 5, 19, 28, 29, and 41—are of a later date. Depictions of donors in the Stein collection paintings illustrate the widespread use of the ancestor portrait format to represent both donors and beneficiaries, chief among them the deceased relatives of the donor. The dating of these paintings may indicate that this genre of portraiture developed in parallel across many media, perhaps sharing a root source in common with the first ancestor portraits of cave 231.

For example, the four donor-worshipper figures of Stein painting 19 (dated 963 CE) kneel in a symmetrical arrangement in a separate register below an image of Bodhisattva Ksitigarbha (fig. 8).[29]fig 8 The object at the center of the group of donors is unmistakably the same sort of inscribed stone tablet that appears in ancestor portraits in caves—its carved base indicates that it is a three-dimensional object as well as a surface for the donor inscription.[30] The inscription beside the second female figure, who holds a plate of offerings, identifies her as a “deceased mother” gumu故母. While the inscription of the man kneeling directly opposite is damaged beyond legibility, his size and placement, which pair him with the mother, would seem to suggest he is the deceased father. According to their inscriptions, the other two kneeling figures seem to have been painted during their lifetimes. Instead of holding offerings like the deceased mother and (presumably) deceased father, they fold their hands in prayer.

In the case of painting 19 — as in Stein paintings 1, 2, 5, 28, 28* (verso) and 29, which include similar configurations of kneeling donor-worshipper figures at their bottom register—the ancestor portrait format can also be seen with dead ancestors and living donor-worshippers sitting side by side. Perhaps because of the spatial constraints of a two-dimensional painting, it seems that the format served a double purpose by portraying both (i) the donors of the painting (to index the donation and place them in the picture as worshippers) and (ii) the deceased parents they donated for (the beneficiaries of the merit generated by creating the painting, who also appropriately worship in the picture). Whether the painting acts as a Buddhist funerary portrait, or simply commemorates the dead beneficiaries of donation as Buddhist convention stipulates, the portrait thus seems to portray a tight-knit family unit, united in life and death under the image.

While these freestanding paintings prove that ancestor portraits were not a genre isolated to wall painting, the architectural placement of ancestor portraits in Buddhist cave temples implies a necessary relationship with a worshipping viewer and therefore inflects the reception of such portraits with religious reverence. In his discussion of a text describing one ninth century miaozhenxiang installed above a temple doorway, “Encomium on the Portrayed Truth of the Head Monk Cao,”[31] Kyan highlights the text’s description of “looking up” zhan瞻 at the portrait of the deceased as the visitor entered and exited the space: this, he argues, refers not only to the physical act of looking but also to a worshipful act of recognition and reverence. Kyan’s argument might be further extended to connect the act of looking up at the portraits with the placement of the portraits directly across from the western Buddha niche. The standard Buddhist practice of clockwise circumambulation allows us to reconstruct the way that a viewer was ideally meant to move about the space, starting with the central Buddha niche then moving around the walls clockwise to encounter the ancestor portraits last before exiting.[32] The movement of the viewer and the act of looking up draw a clear parallel between worship of the Buddha and acknowledgement of the ancestors. In practice, then, the western Buddha niche and the ancestor portraits bookend a ritualized viewing experience as two landmarks in the cave space.

In addition to the ancestor portrait’s placement above the doorway on the east wall, the way in which its pictorial space mirrors the shape of the Buddha niche raises the parents up to receive worshipful address even as it positions the ancestor figures so that they can view and worship the Buddha icon directly opposite them. The west wall Buddha niche of cave 231 is a deep, right-angled rectangular alcove with a horseshoe-shaped raised platform on which the Buddha and surrounding bodhisattva sculptures rest on carved bases. In the ancestor portrait directly opposite, on the east wall, the bounded area of the image, volumetric rendering, and spatial recession create the illusion of a similarly shaped niche. These are two “niche” spaces for worship, one real and the other represented, and they are linked along the central axis of the cave. This axis runs the length of the corridor; were one to draw it, it would connect directly to the Buddha sculpture standing at the center of the western wall niche flanked by a symmetrical arrangement of sculpted bodhisattvas (fig. 9).fig 9 In addition to serving as the axis of architectural symmetry, this axis also parallels the sight lines of a visitor entering or exiting into the cave through the corridor. Significantly, the ancestor portraits sit approximately at Buddha-eye level. The implications of this are twofold. First, because the figures in the portrait occupy a prominent position, with their implied reverence of the icon directly before them, they seem to sit in the presence of the Buddha in perpetuity. Second, while from one perspective these figures worship the Buddha, from another, their ancestor portrait also mirrors the Buddha icon’s function as an object for the viewer’s worship precisely because it is the spatial counterpart to the central Buddha icon.

The possible range of interpretations of the spatial parallel between the Buddha niche and the ancestor portrait outlined above, as well as the iconographic content of the portrait, invite some key questions about what precisely the ancestor portrait is meant to represent. Can we say with certainty that the ancestors are worshipping the Buddha? If so, are they also worshipping the tablet symbol of themselves simultaneously? While it is not possible to answer these questions in definitive terms, I believe that they are also not mutually exclusive. One can easily imagine the double act of paying homage to the Buddha and paying homage to these ancestral images—all in the simple gesture of looking up.

This theory would indicate that the function of ancestor portraits as icons is embedded in their very composition. In other words, unlike standard donor-worshipper images, ancestor portraits exhibit features of iconic representation. In his analysis of Han Dynasty gable carvings of the Queen Mother of the West in The Wu Liang Shrine, Wu Hung defines an iconic representation as an image in which “the central figure exists not only in the pictorial context within the composition but the significance of its representation also relies on the existence of a viewer outside it.”[33] The icon faces out of the composition, establishing a “direct relationship between the icon and the viewer”. Balanced around the central tablet (if not always perfectly symmetrically), the positions of the figures in the ancestor portraits guide the viewer’s focus toward the central tablet with their kneeling postures. In this sense, it is the tablet—the center of focus that faces the viewer squarely—that constitutes the iconic center of the composition.

Applying Wu’s distinction between a viewer who witnesses and a viewer who becomes a participant, I argue that the primary relationship in portraits of deceased ancestors, unlike that in standard donor-worshipper images, is the one between viewer and image. Wu suggests that the particular effect of iconic composition is its ability to transcend both the image’s temporal moment and the pictorial space that it represents; it is in this sense that the iconic composition interpellates the viewer. Certainly, we might imagine this to be a specific viewer, perhaps a family member visiting cave 231 in the years after its completion or generations later to worship in the family’s own space. However, imagining a specific viewer is effectively ancillary to the fact that the concept of the viewer as a participant is already embedded in the form of the iconic image.  The very structure of the ancestor portraits locates the viewer in an act of recognition that, because of their architectural placement and the identity of their figures, is also reverential.

This act of worship might be conceived of as another strand in a web of worshipping relationships; the ancestor portrait forms a pivot, an object of reverence for the viewer that also represents figures who are themselves worshipping the Buddha. Under this analysis, the Buddha would act as the central focus of worship for the viewer. All architectural and decorative elements from the donors in the bottom register to the lateral symmetry of the wall paintings would therefore seem to reflect around its axis of symmetry and direct focus to its niche, in confirmation of its paramount importance. But, like all icons, the Buddha’s presence radiates beyond its immediate context and compositionally acknowledges the viewer’s presence, situating him or her within the space and thus completing this web of relationships. This set of relationships between the Buddha, the ancestor portrait, and the cave space thus encapsulate the viewer, and the various strands weave together to tie him or her all the more tightly to the other world that the chamber represents.


IV. Variation within the ancestor portrait genre

         Caves with ancestor portraits at Mogao number around a dozen, and while a detailed discussion of each portrait exceeds the scope of this study, I nonetheless should highlight select examples of portraits from caves other than cave 231. On the one hand, because ancestor portraits are so consistent in composition, it is possible to consider the portrait of Yin Jiazheng’s parents a definitive representative of the genre; and yet, on the other, each is so unique in its particular handling of the figures that without exploring the diversity of the genre one runs the risk of overlooking the high degree of importance that the genre places on individualization. The ancestor portrait is not only a carefully composed icon; it is also a set of pictorial conventions that could be adapted to the needs of different patrons and different spaces. Examining different ancestor portraits confirms that the features they consistently hold in common—their balanced, centrally-focused composition and their placement above the doorway on the east wall—are key to the relationship they establish with the worshipping viewer. When such information is forthcoming, it is also possible to demonstrate that the portraits always represent deceased close kin of the cave patron. This shows that ancestor portraits invariably elevate and commemorate close kin who are both the beneficiaries of Buddhist donation and those specific family members entitled to filial reverence. Because of their power to connect directly with the viewer, the portraits link the viewer, in the present, to the familial authority of the past. Consequently, they implicate the viewer in a hierarchy of worship constructed within the cave space.

The ancestor portraits of caves 231, 144, 12, 20, and 359 all depict the mother and father of the cave patron, and while the structure of these portraits is highly standardized, representations of each deceased parent were executed with a high degree of individualization (fig. 14). In cave 12, dated to 869 CE, two figures who are likely the parents of cave patron Suo Yibian索義辯 appear above the east wall doorway kneeling on raised platforms elaborately carved like rich domestic furniture around a central tablet with a decorated cap identical to that in the ancestor portrait of cave 231. While the fathers in caves 20 and 144 wear the same red official robe as Yin Bolun in cave 231, Yin Bolun’s form is stocky and solid; the portrait of the father in cave 20, in contrast, represents him with a lanky figure and a long neck. The father in the portrait in 144 sits firm, with a straight back and a neck covered by a thick white beard; in cave 359, alternately, the father is clearly a portly gentleman with a clean-shaven double chin. Likewise, the costumes of the mother figures, such as those in cave 20 and cave 12, are each unique but also elaborate in detailed brocades, transparent silks and jewelry, indicating the specific wealth and status of each. Clearly, these portraits could not have been stamped from a premade stencil, as standard donor-worshipper images were—for in each case, while the painter may have obeyed a set of conventions, he customized each portrait to a significant degree.

Interesting variations on the mother and father ancestor portrait structure can be found in caves 138 and 9.[34] The ancestor portrait over the east doorway of cave 138 also depicts figures kneeling on raised platforms around a central tablet, flanked by groups of attendants, but instead of two kneeling figures, there are three: two female and one male, all holding incense censers. Notably, this portrait better articulates the surrounding space, which resembles a richly decorated private interior—the low platforms with carved frames and decorative floor coverings are invested with more than the usual amount of detail, matched by large groups of male and female attendants at each side[35]. Paired with the tablet at center, one male and one female figure maintain the visual symmetry of the composition, while on one side four male attendants seem to counterbalance the additional woman. The ancestor portrait on the east wall of cave 9, dated to 892 CE, also privileges compositional balance, showing four male ancestors seated on raised platforms around the center pillar, their shared oval eye-shape seeming to underscore the familial bond between them.

The ancestor portrait on the east wall of cave 156, a cave famous for the intricate processional paintings of the victorious Zhang Yichao and his military entourage that line the bottom register of the main chamber, presents a much more enigmatic deviation from the standard ancestor portrait format (fig. 10). While its architectural position, decorative rectangular frame, and deep niche-like pictorial space clearly identify it as an ancestor portrait, instead of figures kneeling around a tablet, the image shows four standing figures, three women and one man, on a decorated carpet-like rectangular surface separated by vertical bands of inscriptions without a tablet. Their postures are those of more traditional donor-worshipper images, standing in three-quarters view, but even without the black lines that originally defined their figures it is still apparent from the shape of their bodies and the curves of their profiles that each was rendered with attention to individual likeness.

Only the leftmost inscription in a long cartouche by the female figure survives. It reads, “The deceased mother called Lady Songguo of the Chen family worships/donates wholeheartedly”[36] Dunhuang Academy scholar He Shizhe 賀世哲 identifies this woman as Lady Chen, the mother of Zhang Yichao張議潮, a local general who ousted the Tibetans and founded the Zhang family regime to rule Dunhuang, who began the construction of cave 156. His nephew and political successor, Zhang Huaishen 張淮深, continued the construction and added to the cave (in his uncle’s honor)[37] in 867 CE, several years after the first set of paintings were complete.[38] While it would be logical for Zhang Huaishen to include his uncle in a space of such memorial importance and architectural prominence—he was, after all, the beneficiary to whom the cave is dedicated—it is impossible to know if these portraits were commissioned by Zhang Yichao to represent his ancestors, or added a few years later by Zhang Huaishen to further reinforce his familial ties to a lineage of political power and demonstrate his reverence for his and his uncle’s ancestors.

While kneeling figures of the deceased are absent from this unique ancestor portrait, kneeling figures at the base of the west wall of cave 156 may provide evidence that certain features of ancestor portraits were not exclusively confined to the upper part of east walls of caves. Remarkably, five figures to the left of the lower register of the west wall below the main niche are represented in the style of ancestor portraits. The figures, who wear monk’s robes and simple red official robes, each kneel on one knee on a raised platform carved with patterns characteristic of domestic furniture. Their hands clasp offerings and they are attended by smaller figures to the side. They are painted with the attention to individual features that is characteristic of ancestor portraits—no two are precisely alike in the drapery of their costumes, the outlines of their figures or the shape of their bodies.

Unfortunately, the inscriptions on these figures are obscured, so we cannot know with certainty whether these portraits represent the deceased. However, their composition—the three-quarters kneeling posture on raised platforms, and the high degree of individualization—makes it possible to venture a guess that the figures represent deceased family members of the Zhang family or a related clan. As contemporaneous paintings in the Stein collections showing deceased kneeling donors confirm, this posture seems to be closely associated with commemorating the dead. Moreover, representing the deceased in the bottom registers of wall paintings arrayed among their living counterparts was common practice in wall paintings from before the ninth century—as suggested before, these images served a commemorative function, though their significance lies in their indexical function rather than any iconographic meaning.


IV. The dead in standard donor-worshipper images

In every cave that contains ancestor portraits, as in virtually every Buddhist cave temple built in the preceding centuries, wall paintings include depictions of the laypeople and clergy who contributed financially or had donations made in their honor. These donor-worshipper images vary widely in size, posture and detail, but a few features distinguish what I will call “standard donor-worshipper images” from other depictions of laypeople. The figures stand in groups, close but not quite touching, as they seemingly process toward the Buddha. They are arrayed according to gender, with women and men lined up on the north and south walls respectively, and their order hierarchically reflects their generation, age, or relationship to the cave patron family. The only individual markers that distinguish these nearly identical figures from others of the same type, replicated from standard forms, are their dress, headgear and accessories.[39] The dead often appear among lines of donors in standard donor portraits, distinguished from their living counterparts only by gu故 or wang亡 prefacing their titles in their inscriptions to designate them as deceased.[40] For example, based on surviving inscriptions, the lower register wrapping under the central platform on west wall of cave 144 includes at least six such deceased mothers, daughters, sons and brothers, intermingled among the living, represented with the same faces, hair, dress, coloring and posture as their living neighbors (fig. 11).

These donor-worshipper figures operate as indexes of contribution and participation—at their most basic level, they could be understood as markers of merit transactions almost like contracts sealing the donation, as Kieschnick describes donor inscriptions.[41] It was immaterial whether the recipient of Buddhist merit was alive or had passed out of this life, like the generations of ancestors who received the benefits of cave building. Moreover, the lack of distinction between living and dead in most of these depictions suggests that a different schema was used to categorize and hierarchize these figures. It seems that in these instances, family, generational hierarchies, and other types of relationships structure representations of laypeople rather than any visual mode of separating living from dead.

Because standard donor figures are modulated variations on stock forms, unlike ancestor portraits they do not represent individuals and they emphatically do not behave as icons. Rather, they are fundamentally tied to their space and derive their significance from their sameness and repetition. Any deceased worshippers were painted to index an act of donation made by a loved one in their honor, and their sameness indicates that there was no perceived need to distinguish between donations for both the living and the dead, since both hope to be reborn in paradise, and both will figuratively worship the Buddha in the cave in perpetuity. The inclusion of the dead among the living might therefore symbolize family cohesion within the sacred space of the family cave, where generations of close family members as well as distant relatives could figuratively occupy the same space. This egalitarian unity between the living and the dead would thus seem to suggest an encyclopedic function served by the family cave: it was a space in which generations of a family and assorted relatives could be indexically gathered together to worship simultaneously, a space where family ties could stretch across temporal boundaries. In this sense, standard donor-worshipper images stress both multiplicity—by metonymically representing the sheer volume of support for the cave, the number of beneficiaries of the act of building—and continuity—by serving as a visual reference to the generations of family members and distant relatives brought together in a single space to worship for all time.


V. Posthumous portraits of politicians: Larger-than-life corridor portraits

Representations of donor-worshipper images in the corridors of ninth and tenth century caves took full advantage of their narrowness, the movement of the viewer, and the ‘transitional’ aspects inherent in the architecture of the corridor. Connecting the main chamber of the cave to the exterior of the cliff face, sometimes passing through an antechamber, corridors bridge the exterior brightness and the gloom of the caves’ interior, guiding the viewer from a clearly lit space to a mysterious interior where the intricate paintings would need to be seen by torchlight. Conceptually, the corridor also marks a transition between outside and inside, from the outside world to the paradisiacal Buddha-land of the interior; just so do contemporary sources describe the interior of caves, imagining them quite literally to be paradisiacal realms that transported donors and visitors. As Schmid has put it, this realm is an “as-if” space of different temporal and cosmic dimensions[42]. Both the size of these images – which are approximately six feet high and placed a foot off the ground – and their arrangement in processional groups amplify a visitor’s sense of the narrowness of the corridor space. They literally stare down over the head of the visitor and appear to stand close by his or her side.[43] If the visitor entered through the passageway, these enormous figures would stand in three-quarters view and face the same direction as he or she does, with the most important figures, often those ones who are portrayed posthumously, heading the procession. Torchlight would illuminate the figures one by one, making the presence of each figure felt individually as the visitor would draw close; likewise, the other figures would fade into darkness as the visitor proceeded. Walking out of the cave chamber, the viewer would face the stares of the painted figures, who would seem to watch with unblinking eyes as he or she exited. From the size and placement of these figures alone, it is clear that they engage with the viewer in a physical and immediate encounter: an imposing dynamic of power and authority seems inherent in their very form.

Larger-than-life donor portraits line the corridors of five caves with ancestor portraits, caves 231, 144, 156, 9 and 12 (see fig. 13 for the portraits in cave 9).fig 13 In each of these caves, where donor inscriptions have survived to identify figures, the men portrayed are given titles that emphasize their authority in Dunhuang society – among them are Dunhuang rulers Cao Yijin, Zhang Yichao, Zhang Huaishen, as well as Guiyijun military and civilian officials, and their relatives. They are arrayed by generation and rank, with the most important leaders such as Zhang Yichao or Cao Yijin (the founders of their respective family reigns) depicted in posthumous portraits leading the processions. Death elevates a figure in the hierarchy; it seems their authority increases with seniority, not least because they represent a link to the past and serve to illustrate the continuity of rule.

Family ties and political authority are historically inseparable as well as conceptually intertwined, especially because members of the ruling Zhang family and the Cao family that succeeded them were personally some of the most important patrons at Dunhuang. Other powerful families with the means to construct family caves were almost invariably related by blood or marriage to the Zhang and Cao clans. Cave 156 exemplifies the case in which a ruler was also the patron. Standing at the western end corridor of cave 156 is a posthumous portrait of Zhang Yichao, the Dunhuang general who freed Dunhuang from the Tibetan occupation and reestablished local rule, standing opposite ladies of the Song family (fig. 12)[44].fig 12 Male relatives, including his nephew and successor Zhang Huaishen, stand in line behind him in the same pose. The air of political authority that the grand titles in their inscriptions would convey to the literate reader compliments the impression that the massive figures make as visitors pass through the corridor. Zhang Yichao’s portrait displays a degree of individualization that is equaled only in ancestor portraits. Though the details of his face have been lost, his wide, stocky frame, round face, short neck and pursed lips seem to painstakingly communicate his likeness to the extent that a posthumous portrait painted nearly sixty years later in cave 98 is nearly identical. Both the portraits in cave 156 and cave 98 depict Zhang Huaishen, holding a round incense censer and standing directly behind Zhang Yichao— in both representations, he physically resembles a smaller Zhang Yichao, broader than surrounding figures and with a wider jaw, second in size to the head of the group.

Occasionally depicted opposite a procession of their wives, these political figures appear even in those caves where the cave patron family’s relationship to the ruling family is established only through marriage, is distant, or is only tangential. From examining the familial and marital relationships between these political figures and the cave families in the handful of caves where inscriptions are still legible, it is clear that political relationships, not family ties, dictate their inclusion in the corridor space. For example, in the tenth century, approximately one century after its initial construction, the corridor of cave 231 was repurposed (or perhaps merely refurbished) to add the images of the contemporary ruler of Dunhuang, Cao Yuanzhong 曹元忠, and officials of the Zhai 翟family.[45] The fact that renovators installed portraits of the contemporary ruler in the corridor of this Yin family cave indicates that the perceived purpose of this corridor space was to represent figures of authority rather than the cave patrons or their close family members. Since the ruling families and the Yin patrons were linked by marriage, there is still an element of familial unity motivating the inclusion of these figures in the corridor space. Nevertheless, the prominence given to distant relations and previous generations clearly privileges the person’s political importance, often synonymous with their place in a direct line of descent, over their relationship to the cave patron’s family.

The deceased figures leading the procession visually represent the generational succession of rulership and suggest that political authority is derived from continuous lineage. Their architectural placement, arrangement in processions, size, and the physical impact that they impart serve together to interpellate the worshipping visitor into a powerful viewing relationship. Because of their visceral impact, the figures have the capacity to impress the viewer with the cave patron family’s ties to political power deeply rooted in generations of history. The corridor space secures the link between the cave interior—the sacred space of a single family, inhabited by generations of worshippers and presided over by the patron’s ancestors—to the sacred site of the Mogao Caves—worked and reworked by successive generations, a site built as a collective display of devotion. But, moreover, because of the corridor’s architectural role as a transitional space, the painted figures direct the viewer, priming the viewer for entry into the main chamber. At the same time as the viewer falls under their sway, he or she emulates their gesture—proceeding forward to worship the Buddha. Once again, a hierarchy of worshipping relationships comes into play, in which the viewer finds him or herself obeying the lead of the painted figures who in turn show their reverence to a higher authority inside the cave. While they radiate their own worldly power — a power derived from their real-life authoritative status and articulated visually in their size, placement and pictorial composition— they are also architecturally and conceptually subordinated to the sacred power in the cave interior. As the viewer passed out of the corridor space and into the main chamber, he or she would carry, in memory, the symbolic power of the paintings encountered on the walls of the corridor – paintings that are both the preface and conclusion to the viewer’s experience.


VI. A complete experience

Entering the cave, the visitor would move through diminishing light past the monumental figures in the corridor, patriarchs of the past and present towering above. He or she would feel them to be physically present on each side, and the literal illustration of a lineage of power that transcends generations would serve to connect, temporally, his or her living body with the dead represented all around. By literally bridging the outer world and sacred space of the cave interior, these corridor portraits thus represent a hierarchy of authority—they situate and dominate the worshipping visitor, but they also stand on the margins of the cave, processing toward the central Buddha icon and worshipping it in perpetuity. The viewer, too, must stand beneath their gaze and, like the painted figures, direct his or her movement and the focus of vision toward the Buddha. Entering the main chamber, the visitor would be overwhelmed by the intricate and colorful paintings that coat the walls and ceilings—especially in flickering firelight, this would truly seem like another world. Familiar markers of domestic luxury abound, from the chuang-shaped western niche to the screen-like partitions of the wall paintings, painted as if framed with carved wood and filled to overflowing with colorful deities and scenery. These intricate sutra paintings, the registers of donors, and the canopy-like painted ceiling would not simply refer to or represent a paradisiacal realm; they would manifest that paradisiacal realm as a substantive reality, placing that viewer somatically within it.

In the bottom registers of the main chamber, the viewer would see rows of small donors figuratively supporting the Buddha niche and paintings on the walls, symbolizing the multiplicity of devotion and the unity of the family, living and deceased, as they worshipped together for all time. After paying homage to the Buddha, as so many painted figures do perpetually, and turning to exit the main chamber, the visitor would look up at the ancestral figures on the east wall doorway. He or she would pay them homage in that act of acknowledgement, responding to the visual echo of the Buddha niche that contains the ancestral figures who preside over the space. They would command an intimate relationship with the worshipping viewer; their presence would impress upon that viewer a reminder that the ancestors are the reason why the cave exists, and that they are the ones most intimately connected to the sacred space. Proceeding out of the cave, the viewer would encounter once more the line of colossal portraits, this time facing him or her directly, and would move past the deceased leaders of generations and toward his or her living descendants as he or she exited. The viewer would thus transition back into his or her own time, moving out toward the governing authorities of everyday worldly life. In this way, a visit to the caves would offer up a dramatic and fully realized experience for a worshipping viewer.

Why patrons and artisans would choose to transform the cave space with such powerful visual illustrations of these timeless ideas seems almost self-evident—the cave space becomes all the more effective as a symbol of family power and as a place of worship if it connects with the viewer both visually and somatically in order to inspire awe. Ultimately, it is the connection between the viewer and the images of laypeople—familiar figures, here inhabiting the margins between scared and secular, living and dead—that draws the viewer in. Ancestor portraits and corridor portraits become anchors that serve to link the viewer’s experience and his or her body to the sacred and paradisiacal space of the cave. They transport the viewer, shaping his or her encounter with the cave space. The cave’s interior is built from the elements of worldly authority—family ties, political rulership, and ancestor worship—but it also transcends the temporal and spatial boundaries of this authority by integrating these elements seamlessly into a sacred and timeless other world.

Author’s note regarding images:
Any images referenced in the body text that have no corresponding figures published here refer to images drawn from a private database that the Dunhuang Academy generously shared with Professor Wu Hung’s students for his research seminar on the Dunhuang caves. This image database was an invaluable resource in my research, especially because it documents the interiors of many of the caves that are not reproduced elsewhere. Since these images were only intended for the use of faculty and students in the course, they are omitted here.
The images reproduced here are drawn from online sources and from the following published print sources:
Ōnuma Sunao 大沼淳. Tonkō sekkutsu 敦煌石窟 [Dunhuang Caves]. 10 vols. Tōkyō 東京: Bunka Gakuen Bunka Shuppankyoku, 2001.
British Museum. The Art of Central Asia: The Stein Collection in the British Museum. 2 vols. Tokyo; New York: Kodansha International, 1983.

Appendix 1 – Caves with ancestor portraits

Cave Date Family cave (patron) Ancestor portraits Larger-than-life donors in corridor?
231 839 CE Yin Jiazheng 陰嘉政 East wall, man and woman kneeling before tablet with inscription Yes – Cao Yuanzhong, Zhai family members on north wall added in 10th century
359 Cut in the Tibetan period, renovated Cao period Possibly Shi 石, a Sogdian family (proposed by Sha Wutian) East wall, woman and man kneeling before wide square tablet without text (extremely short corridor out of structural necessity)
156 865 CE Zhang 張– started by Zhang Yichao 張議潮, completed by nephew  Zhang Huaishen 張淮深, dedicated to Zhang Yichao East wall, woman and three men standing, no tablet.West wall, under dais, group of kneeling figures with damaged inscriptions Yes – Zhang Yichao, Zhang Huaishen
12 869 CE Suo Yibian 索議辯 East wall, woman and man kneeling before tablet with inscription Yes – inscriptions too damaged to identify figures
20 Zhang period (845-906 CE) (unknown) East wall, woman and man kneeling before tablet without text (corridor collapsed)
144 Zhang – Cao periods (paintings) Suo 索 East wall, woman and man kneeling before tablet with inscription Yes – inscriptions too damaged to identify figures
138 Ca. 890 – 910 CE, opinions differ Yin陰 – exact patron unknown, most likely a son of Yin Lifeng陰李丰, different branch of family and one generation after cave 231’s Yin patrons East wall, woman and man kneeling before tablet with text, group of attendants No – repainted in the Yuan with larger than life standing bodhisattvas
9 890-893 CE (892 acc. to Ma De) Zhang Chengfeng 張承丰 East wall, four men kneeling before tablet without text Yes – Guiyijun Official Suo, Li Hongjian, Provincial official Zhang, another Li, etc.
420 Sui cave, ancestor portrait on east wall added in Cao period (unknown) East wall, placed below Manjusri image (slightly overlapping), man and woman kneeling before square tablet with no text No – repainted in Cao Guiyijun period with larger than life standing bodhisattvas





British Museum. The Art of Central Asia: The Stein Collection in the British Museum. 2 vols. Tokyo; New York: Kodansha International, 1983.


Dunhuang wen wu yan jiu suo 敦煌文物研究所. Dunhuang Mogaoku Neirong Zonglu 敦煌莫高窟內容總錄 [Comprehensive Record of the Contents of the Dunhuang Mogao Caves]. Beijing: Wen wu chu ban she, 1982.


Dunhuang yan jiu yuan 敦煌研究院. Dunhuang Mogaoku Gongyang Ren Tiji 敦煌莫高窟供養人題記 [Donor Inscriptions of the Dunhuang Mogao Caves]. Beijing: Wen wu chu ban she, 1986.


Foong, Ping. “Multipanel Landscape Screens as Spatial Simulacra at the Mogao Caves, Dunhuang,” in Essays on East Asian Art In Honor of Professor Wen C. Fong, ed. Jerome Silbergeld et al., vol. 2, 2 vols. (Princeton: P.Y. and Kinmay W. Tang Center for East Asian Art; Department of Art and Archaeology, Princeton University, 2011), 533–556.


Fraser, Sarah Elizabeth. Performing the Visual: the Practice of Buddhist Wall Painting in China and Central Asia, 618-960. Stanford, Calif.: Stanford University Press, 2004.


———. “An Introduction to the Material Culture of Dunhuang Buddhism: Putting the Object in Its Place.” In Asia Major 17, no. 1 (2005): 1-14.


He Shizhe 賀世哲. “從供養人題記看莫高窟部分洞窟德營建年代 [Regarding the Periodization of Some Mogao Caves from the Perspective of Donor Inscriptions].” In Dunhuang Mogaoku Gongyang Ren Tiji 敦煌莫高窟供養人題記 [Donor Inscriptions of the Dunhuang Mogao Caves]. Beijing: Wen wu chu ban she, 1986.


Kyan, Winston. “Family Space: Buddhist Materiality and Ancestral Fashioning in Mogao Cave 231.” The Art Bulletin XCII, no. 1–2 (June 2010): 61–82.


———. “The Body and the Family: Filial Piety and Buddhist Art in Late Medieval China” Dissertation. University of Chicago, 2006.


Lee, Sonya. “Repository of Ingenuity: Cave 61 and Artistic Appropriation in Tenth-Century Dunhuang.” The Art Bulletin XCIV, no. 2 (June 2012): 199–225.


Ma De 馬德. Dunhuang Mogaoku Shi Yanjiu 敦煌莫高窟史硏究 [Research on the History of Dunhuang Mogao Caves]. Di 1 ban. Lanzhou 蘭州: Gansu jiao yu chu ban she, 1996.


McNair, Amy. Donors of Longmen: Faith, Politics, and Patronage in Medieval Chinese Buddhist Sculpture. Honolulu: University of Hawai’i Press, 2007.


Ning Qiang. Art, Religion, and Politics in Medieval China: The Dunhuang Cave of the Zhai Family. Honolulu: University of Hawai’i Press, 2006.


Schmid, Neil. “Material Culture in Dunhuang.” Asia Major 19, no. 1–2 (2006): 171–210.


Sha Wutian 沙武田. “吐蕃統治時期敦煌石窟供養人畫像考察[A Study of Donor Portraits in Tibetan Period Dunhuang Mogao Caves].” 敦煌學輯刊 [Journal of Dunhuang Studies] 62, no. 2 (2003): 80–85.


———. “莫高窟第138 窟智惠性供養像及相關問題研究[Research on the Zhihuixing Donor Portrait of Mogao Cave 138 and Related Questions].” 敦煌學輯刊 [Journal of Dunhuang Studies] 53, no. 3 (2006): 83–89.


Sha Wutian 沙武田, and Bai Tianyou 百天佑. “莫高窟第231窟陰伯倫夫婦供養像解析 [A Study of the Donor Portrait of Yin Bolun in Mogao Cave 231].” 敦煌學輯刊 [Journal of Dunhuang Studies] 96, no. 2 (2006): 6–10.


Waley, Arthur, British Museum Dept. of Prints and Drawings, and Delhi (India) Central Asian Antiquities Museum. A Catalogue of Paintings Recovered from Tun-huang by Sir Aurel Stein, K.C.I.E., Preserved in the Sub-department of Oriental Prints and Drawings in the British Museum, and in the Museum of Central Asian Antiquities, Delhi; London: British Museum, 1931.


Whitfield, Roderick, and British Museum. The Art of Central Asia : the Stein Collection in the British Museum. Tokyo; New York: Kodansha International, 1983.


Wiercimok, Edith. “The Donor Figure in the Buddhist Painting of Dunhuang.” Silk Road Art and Archeology 1, no. 1 (1990).


Wu Hung. The Wu Liang Shrine: The Ideology of Early Chinese Pictorial Art. Stanford: Stanford University Press, 1989.


———. “What Is Bianxiang? On the Relationship Between Dunhuang Art and Dunhuang Literature.” Harvard Journal of Asiatic Studies 52, no. 1 (June 1992): 111–192.


Zhang Jingfeng 張景峰, and Gu Shuyan 顾淑彦. “敦煌莫高窟第138窟供養人畫像再認識 [Further research on the donor portraits in Cave 138 at Mogao Grottoes in Dunhuang].” Hundred Schools In Arts 藝術百家 108, no. 3 (2009): 17-23.


Zheng Binglin 郑炳林. Dunhuang Bei Ming Zan Ji Shi 敦煌碑銘贊輯釋. Lanzhou蘭 州: Gansu jiao yu chu ban she, 1992.

[1] The Chinese term for donor-worshipper image, gongyang ren xiang 供養人像, conveys the meaning of both “donation” and “worship”, concisely describing these two important functions of the images in a single term. In English, often these portraits are simply called “donor images,” a term that draws an apt parallel with portraits of donors painted into Christian religious art.

[2] See, for example, Dorothy C. Wong, Chinese Steles: pre-Buddhist and Buddhist use of a symbolic form (Honolulu: University of Hawai’I Press, 2004) and concerning posthumous image of the late Northern Wei emperor in the Binyang Central Grotto imperial procession sculptures, see Amy McNair, “The Mechanics of a Karmic Gift of Sculpture” in Donors of Longmen, 31-50, especially pages 43-44.

[3] According to Ma De’s 馬德 dates.

[4] Please see Appendix 1 for a chart detailing the dating and patronage of caves containing ancestor portraits.

[5] I am using “standard donor-worshipper image” to denote generic portraits of worshipping figures typically arrayed in rows in the bottom registers of the main cave chambers. These were stamped out from standard templates when they were painted, so that while the size and details of each figure’s costume and accessories may vary, groups of figures are identical in body shape and facial structure.

[6] The distinction of where sacred power inheres—in the thing itself, rather than just in the concept that the thing evokes— is subtle but crucial. While it may be impossible to know for sure how contemporary viewers experienced these religious spaces, there is ample physical and textual evidence of the sacred power of physical objects and spaces in Buddhist belief. See John Kieschnick, “Sacred Power” in The Impact of Buddhism on Chinese Material Culture (Princeton: Princeton University Press, 2003) 24-82.

[7] Sarah Fraser has commented on the “horror vacua”, an aversion to leaving empty space in artistic composition, evident in Dunhuang wall paintings. She implies horror vacua is part of a strategy to render the intangible tangible and to lend a material tactility to the imaginary. Sarah Fraser, “An Introduction to the Material Culture of Dunhuang Buddhism,” 8.

[8] Winston Kyan, “Family Space and Ancestral Fashioning,” 61.

[9] Schmid, “Material Culture” 172, 208-209.

[10] Ping Foong, “Multipanel Landscape Screens,”

[11] Ibid., 540.

[12] By my calculations based on Ma De’s dating and periodization scheme outlined in 敦煌莫高窟史研究 (Research on the History of the Dunhuang Mogao Caves), (Beijing: Wen wu chu ban she, 1996), the rate of cave building during the Zhang Guiyijun period (850-914 CE) averaged 1.25 caves per year, outstripped only by the Sui Dynasty when an average of 2.38 caves were built per year. The Cao Guiyijun period (914-1036 CE) saw a sharp decline in new cave building, down to approximately one new cave every three years on average, but the rate of renovation increased dramatically to 2.38 caves renovated per year. Some of these renovations were more comprehensive than others, but many renovations completely remade older caves to the point that they are considered Cao Guiyijun period caves in style.

[13] Kieschnick, 198.

[14] Two manuscripts record the contents of the stele, called the Yin Chushi Bei, 陰處士碑 for short: P.  4638 and P. 530. Of these P. 4638 is the most complete text and the one that I will reference here using Zheng Binglin’s transcription. A long paper manuscript inscribed on both sides with many different writings, P. 4638 can be found online at;recnum=61932;index=2.

Both versions of the Yin Chushi Bei are transcribed in Zheng Binglin 鄭炳林, Dunhuang Beimingzan jishi 敦煌碑銘贊輯釋, (Lanzhou 兰州: Gansu jiao yu chu ban she, 1992).

[15] Chinese transcription in Zheng, Dunhuang Beimingzan jishi, 240. English translation my own.

[16] English translation my own, with many thanks to Paul Copp for his invaluable advice on translating this text.

[17] Sha Wutian interprets the jun 君, the lord/ruler, in this phrase, baoen jun qin 報恩君親, as a subtle reference to the actual rulers of Dunhuang at the time, the Tibetans, to support a broader iconographic reading of the ancestor portrait of cave 231 as a symbol of reconciliation between the filial piety of traditional Han Chinese culture and the Tibetan occupation. However tenable his interpretation may be with the support of other types of evidence, this phrase bao’en jun qin is such a stock dedicatory phrase, one that appears in countless Buddhist inscriptions made before the ninth century, that it seems something of a stretch to interpret this character jun君 “lord/rulers” as a direct reference to particular rulers.

[18] Kyan, “Body and the Family.”

[19] See Appendix 1, a table with dates of construction, patrons and descriptions of ancestor portraits in each cave. I have identified Caves 231, 144, 138, 359, 156, 9, 12, 20 and 420 as caves containing ancestor portraits. 143 and 238 also contain ancestor portraits according to Sha Wutian– in the latter case, the space on the east wall is now too smoke-damaged to confirm this.

[20] Inscription transcribed in Dunhuang yanjiuyuan 敦煌研究院, Dunhuang Mogaoku gongyangren tiji 敦煌莫高窟供养人题记 [Donor Inscriptions of the Dunhuang Mogao Caves], (Beijing: Wen wu chu ban she, 1986), 105. English translation my own.

[21] Winston Kyan, “Family Space: Buddhist Materiality and Ancestral Fashioning in Mogao Cave 231,” 72. Also discussed in Sha Wutian 沙武田, and Bai Tianyou 百天佑, “莫高窟第231窟陰伯倫夫妇供養像解析 [A Study of the Donor Portrait of Yin Bolun in Mogao Cave 231].”

[22] Winston Kyan, “The Body and the Family: Filial Piety and Buddhist Art in Late Medieval China”, 248.

[23] Kyan, “Body and the Family” and “Family Space and Ancestral Fashioning,” 77. Kyan does explain that ancestor portraits were portrait type that appears in several other caves, though his discussion focuses mainly on cave 231.

[24] Sha Wutian 沙武田, and Bai Tianyou 百天佑, “莫高窟第231窟陰伯倫夫妇供養像解析 [A Study of the Donor Portrait of Yin Bolun in Mogao Cave 231].”

[25] Sha Wutian 沙武田. “莫高窟第 138 窟智惠性供養像及相關問題研究 [Research on the Zhihuixing Donor Portrait of Mogao Cave 138 and Related Questions].”

Zhang Jingfeng 張景峰, and Gu Shuyan 顾淑彦, “敦煌莫高窟第138窟供養人畫像再認識 [Further research on the donor portraits in Cave 138 at Mogao Grottoes in Dunhuang].”

[26] Sha Wutian voiced a similar view on the overall lack of deep inquiry into the visual characteristics of donor-worshipper images, to open discussion of the ancestor portrait of cave 231. “莫高窟第231窟陰伯倫夫婦供養像解析[A Study of the Donor Portrait of Yin Bolun in Mogao Cave 231],” 6.

[27] Zheng Binglin 鄭炳林, Dunhuang Beimingzan jishi 敦煌碑铭赞輯释, 7-10.

[28] Inscriptions on Stein painting 17 are transcribed and translated in Arthur Waley et al., A Catalogue of Paintings Recovered from Tun-huang by Sir Aurel Stein (London: British Museum, 1931) 26-29.

[29] A catalog entry for Stein painting 19 and translations of inscriptions can be found in British Museum, The Art of Central Asia: The Stein Collection in the British Museum (Tokyo; New York: Kodansha International, 1983) 2: 319.

[30] The inscription on this tablet is still legible, and contains a fairly standard dedication with the exception that it identifies the subject of the painting as the cult figure bodhisattva Ksitigarbha, who gained popularity in the tenth century. One part reads, “Also he [the donor] desires that his parents and relations by marriage and all his connections may rest in health and security and that his brothers, cousins and that all the collateral branches of his family may be moistened with the dew of a prosperous position.” Translation in The Art of Central Asia, 2: 319, adapted from Waley, A Catalogue of Paintings Recovered from Tun-huang by Sir Aurel Stein.

[31] Document P. 4660, “都僧政曹僧邈真赞” in Zheng, Dunhuang beimingzan jishi. Partial translation and translation of zhan as “looking up” in Kyan, “Body and the Family”, 244.

[32] Kyan reconstructs a somewhat different sequence of movement in “Family Space”, 77, based on the sequence in which the merit record of Yin Jiazheng lists the contents of the cave’s sutra paintings, which enumerates the paintings’ subjects in a half-counterclockwise (on the south wall) and half-clockwise (on the north wall) sequence. However, it seems that one direction of movement can be counted on, and that is the standard clockwise direction of circumambulation. In either case, the viewer would face the Buddha first, and exit facing the ancestor portraits last.

[33] Wu Hung. The Wu Liang Shrine: The Ideology of Early Chinese Pictorial Art (Stanford: Stanford University Press, 1989), p. 133-136. Wu Hung finds definitive links between these crucial first instances of iconic composition of the Queen Mother of the West 西王母 in Chinese Pictorial art in the 2nd century CE and the earliest appearance of Buddhist icons in China, imported from India during the same time. The fact that in the Han dynasty, Buddhist modes of representation already began to redefine the portrayal of native Chinese deities seems especially relevant here. Buddhist ways of making images can clearly translate into diverse other contexts.

[34] Opinions differ on when cave 138 was completed precisely, ranging from as early as 890 CE to between 900 and 910 CE in the opinion of Zhang Jingfeng, which in combination with successive renovations through the tenth century make it difficult to identify the precise cave patron. However, it is certainly a Yin family cave, belonging to a different branch of the same Yin family of Yin Jiazheng, patron of cave 231.

[35] Kyan, “Family Space”, 68-9.

[36]“亡母贈宋國太夫陳氏一心供養.” Transcribed in Dunhuang Mogaoku gongyangren tiji, 73.

[37] Sonya Lee, “Repository of Ingenuity,” 202.

[38] He Shizhe identifies the male figures, whose inscriptions are now missing, in the portrait from left to right as Zhang Yichao’s father Zhang Qianyi 張谦逸, Zhang Yichao’s brother, and last but not least, Zhang Yichao himself at right. While it’s easy to imagine that the male figure paired with the mother of Zhang Yichao should logically be his deceased father, lacking inscriptions, it is more difficult to prove that the portrait includes Zhang Yichao himself. He Shizhe 賀世哲, “從供養人題記看莫高窟部分洞窟德營建年代 [Regarding the Periodization of Some Mogao Caves from the Perspective of Donor Inscriptions],” In Dunhuang Mogaoku gongyangren tiji,, 209.

[39] Sarah Elizabeth Fraser, Performing the Visual : the Practice of Buddhist Wall Painting in China and Central Asia, 618-960. (Stanford: Stanford University Press, 2004), 27.

[40] Edith Wiercimok, “The Donor Figure in the Buddhist Painting of Dunhuang.” Silk Road Art and Archeology 1 (1990), 206.

[41] Kieschnick, Impact of Buddhism, 163.

[42] Schmid, “Material Culture”, 208.

[43] Size is my best guess based on the measurements of these caves given in Mogaoku gongyangren tiji.

[44] Dunhuang Mogaoku gongyangren tiji, 73.

[45] Dunhuang Mogaoku neirong zonglu, 80.

Comments are closed.