A Mobile Temple: Forms and Visual Grammar of Portable Buddhist Shrines from the 3rd to the 8th Centuries Unearthed Along the Silk Road

Abstract

Portable Buddhist shrines refer to small-scale mobile or assembled shrines, typically made of wood, stone, clay, and metal. They were initially used as temporary ritual sites or ornamental attachments for temples and stupas, later becoming independent objects of devotion. This art form, the origins of which can be traced to ancient India and later diverse regional traditions, has been discovered in significant quantities along the Silk Road and neighboring regions. Previously, scholarly attention centered primarily on exquisite wall shrines, stupa-shaped shrines, and stele-shaped shrines. However, when factors such as the spatial arrangement and ritual functions of mobile ritual sites are taken into account, along with the materials and techniques employed in creating Buddhist shrines, artifacts such as badge-style bronze Buddha statues, painted silk banners, and wooden panel paintings may also be classified as portable Buddhist shrines. Accordingly, portable Buddhist shrines can be divided into three forms: pedestal, hanging, and open–close or mother–child. A key reason for this expanded classification is that all such forms are functionally and stylistically linked to large-scale cave temples. Moreover, these shrines share a common visual grammar, defined by the dynamic integration of images and texts and the mutual imitation and complementarity of statue and painting. This represents a quintessential example of cross-cultural dissemination and the coexistence of local traditions in Buddhist art.

1. Portable Buddhist Shrines and Mobile Ritual Sites

Portable Buddhist shrines refer to objects for Buddhist worship that can be easily moved and assembled. One shrine was structured around a Buddha image, with a pedestal or a backscreen representing the ritual site of Buddha. Crafted from materials such as wood, stone, ivory, clay, and metal, these shrines often served as independent sacred objects. They were used to construct small or temporary altars, accommodating the daily rituals and devotional practices of devotees. Since the 20th century, small Buddhist shrines have been unearthed in the birthplace of Buddhism, along the Silk Road, and in neighboring regions that have continuously expanded our understanding of the types, styles, and socio-cultural functions of Buddhist art from the 1st to the 13th centuries. Previously, scholars primarily employed iconographic and stylistic analyses, based on typical archaeological finds from Central Asia and Xinjiang and Gansu of China. They examined questions regarding the international style, figure representation, visual motif, and narrative structure of portable shrines (Soper 1965; Lee 1966; Granoff 1968; Pal 2004). In recent years, the social–historical approach has gained momentum. Studies on pilgrims (Wong 2018) and the cross-regional transmission of portable shrines have prompted a re-examination of concepts such as “monumentality” and “site-specificity” in Buddhist art (Chan 2023, p. 3). When we turn to specific objects and their transmission, however, the interpretation of portable Buddhist shrines often remains fragmentary. This failure to shift focus from large, fixed ritual centers to peripheral, mobile sites prevents us from connecting various small statues, wooden panel paintings, painted silk banners, and statue inscriptions, connections that would reveal the meaning of objects made to be moved and re-assembled. This paper takes artifacts from the 3rd to the 8th centuries unearthed along the Silk Road and in neighboring regions as case studies. Taking an approach of visual anthropology, this study aims to re-integrate objects fragmented across modern disciplines and examine three questions: How do we define portable Buddhist shrines? What forms emerge from the artifacts unearthed along the Silk Road and neighboring regions? What local traditions and distinct visual grammar do these forms embody?

In terms of social structure, the development of Buddhist art along the Silk Road was sustained by two types of ritual sites. The first type comprised large and medium-scale cave temples, which provided a space for the communities of monks, nuns, and devotees; possessed sutras and ritual implements; enjoyed official patronage; and, consequently, underwent sustainable development. The second type encompassed small-scale, mobile or temporary, and self-organized ritual sites established by devotees, including family temples, Buddhist communities and associations, and private monasteries (Meng 2009; Du 2010; Hao 2019). The two types of ritual sites formed a dynamic, complementary relationship. The large and medium-scale sites commanded geographic, economic, and cultural resources, functioning as the “receiver” and “transmitter” of Buddhist art. Given their scattered distribution, however, large and medium-scale sites could not meet daily religious needs. As a result, Buddhists relied on small-scale sites organized by families or local communities. Through activities such as copying and reciting sutras, constructing shrines, carving statues, gathering devotees, and performing rituals, these sites served to consolidate communal memory, settle the souls of the deceased, or convey the public’s practical appeals for blessings and disaster aversion (Jiang 2008; Shaw 2016; Lin and Xie 2025). In this way, the small sites constituted a capillary network that mediated and sustained the system of large and medium-scale cave temples, dynamically interacting with and reshaping them.

The construction of small ritual sites relied on portable Buddhist shrines. It was precisely these symbolic objects that enabled Buddhist temples to be portable and in any location demanded by ritual practices: atop a person’s head or over the heart; in the main position inside a house; on altar tables, walls, or objects; in eaves and corridors; and in secluded valleys. Through devotional rituals, these shrines could also be installed for display or stored in temple caves and could even be buried in underground tombs. Thus, these shrines not only encapsulated the presence of Buddha, the appeals of devotees, and the fabric of daily life but also reflected the concepts and forms of Buddhist art from large ritual sites, both near and far. Therefore, it is necessary to re-examine the boundary of portable Buddhist shrines, identify their related types, and outline their formal evolution.

To begin with, this paper reviews two questions. The first question is about the early forms of Buddhist shrines. Archaeological excavations at Swat (Pakistan), Bairat (India), and Lumbini (Nepal) reveal that early Buddhist shrines evolved from stupas. Initially, sacred ritual spaces were demarcated using holy trees, wooden pillars, railings, etc. Later, perishable materials were replaced by brick and stone. Cave temples with arched entrances and the Sanchi-Type with domes, walls, and railings emerged (Piggott 1943; Coningham et al. 2013). Notably, in Sanskrit, stupas also signify boxes or containers used to hold sacred bones and relics (Monier-Williams 1960). Archaeological evidence indicates that small Buddhist shrines proliferated around the 1st century. These included architectural decorative components (clay or stone) and independent bronze shrines (Musée des Arts Asiatiques Guimet 2010; Bhandre 2018; Rienjang and Stewart 2019; Hameed and Bukhari 2021). They were all potentially connected to large temples. Functionally, they were offerings from devotees; formally, they were miniature representations of large temple architectures. Therefore, portable Buddhist shrines needed to possess two basic elements: a Buddha image and a spatial sign of Buddha’s ritual site (e.g., arches, entrances, railings, eaves, curtains, steps, pedestals, backscreens, bodhi trees). In the minimalist mode, a single Buddha image and a spatial sign (specific or abstract) constituted a small shrine. In a typical or composite mode, several Buddha images and multiple spatial signs may be included. The two cases unearthed in Pakistan (Figure 1) both involve a comprehensive presentation of the three-disc stupa. However, we noticed that, as architectural components, case (a) only marks the ritual space with bodhi leaves, rose decorations, and railings, while the Buddha statue is omitted (or supplemented by human figures lost on either side). In contrast, case (b) features portable shrines, which incorporate both elements. The two artifacts of similar size also remind us of the flexible development of portable shrines in terms of spatial function, decorative elements, and craftsmanship.

The second question refers to the understanding of portability. Literally, this refers to an object that can be moved or carried. With regard to fixtures in cave temples, objects that require one or more people to move may be broadly classified as portable objects. In terms of production, portability relates to materials, techniques, and donors, as well as cultural traditions. There are materials ranging from the lightest silk and linen to mid-weight wood, plaster, and clay and the heaviest stone and metal—each category possesses distinct properties and expressive potential. The choice of form depended not only on the financial and human resources available to monks, nuns, and devotees but, more fundamentally, also on their conception of what made an object valuable or sacred. Although wood and silk are far less common in archaeological sites due to decay, they—alongside the more frequently found metal, stone, and clay—are equally vital in constituting ritual spaces and demand integrated analysis.

2. The Forms of Portable Buddhist Shrines

Based on the discussion above, this paper proposes classifying portable Buddhist shrines into three styles according to their physical forms and modes of use.

2.1. Pedestal-Style Shrine

As the most common type, the pedestal-style shrine can be divided into the minimalist mode and the typical mode. The minimalist mode generally consists of a Buddha image and a pedestal, sometimes supplemented with a backscreen. A pedestal-style shrine can be placed independently on a surface. Owing to the absence of a complete stupa- or cave-like structure, it is commonly categorized under the broad term “Buddha image”. Material evidence indicates that a pedestal-style shrine was produced through two methods: entire casting and assembly after separate casting. Numerous examples have been found among small wooden, bronze, and plaster shrines excavated in Kucha, Khotan, Turfan, and Dunhuang, as well as among white-clay Buddha images unearthed in Boxing County, Shandong. Many lack their original pedestals but retain tenons for attachment (Figure 2). A gilded bronze standing Buddha unearthed from South Korea, dating to the 7th or the 8th century and now housed in the Asian Art Museum of San Francisco (accession no. B60B37+), also adopts this form.

A typical pedestal-style shrine is a miniature representation of a Buddhist cave or temple. This form crystallizes into two basic types: the cave-style shrine and the house-style shrine. The interior of the shrine is carved with a seated Buddha and Bodhisattva, and its eaves are ornamented with celestial musicians. On the exterior, Vajrapāṇi, auspicious beasts, and donors are carved [as clearly seen in Figure 3d]. The cave-style shrine and the house-style shrine share a spatial layout, demarcated by the shrine entrance, with a clear distinction between the visual center and periphery for Buddha, Bodhisattva, and donors inside and outside, embodying a linear narrative structure. Building upon the cave-style and house-style shrines, the addition of ornamentation and the combination of modular units form two types of portable pedestal-style Buddhist shrines: the stupa-shaped shrine and the stele-shaped shrine, both of which exist in vast quantities.

The stupa-shaped shrine features at least four facades and can be expanded vertically by adding decoration units. With its multi-directional viewpoints—upper-lower, front-back, and left-right—it embodies a multi-layered visual structure. The stupa-shaped shrine varies in form (circular or square) and construction material (brick, stone, wood, etc.). These combinations accordingly result in diverse manifestations. For example, the Five-Level and Four-Dimensional Pagoda unearthed in Dunhuang was modeled on stone stupas from the 4th to the 5th centuries. Each facade contains a shrine with a relief-carved, seated Buddha flanked by two donors. The presence of ink outlines on its caisson ceiling and around the shrine [Figure 3a] provides physical evidence for the extended period of use. The stone stupa on a vajra base unearthed in Datong was modeled on a wooden pavilion, with the number of Buddhist shrines and Buddha images on each tier varying according to the architectural layout [Figure 3b]. The form of these two stupa-shaped shrines accords with contemporaneous pagodas in Dunhuang and Pingcheng, yet differs from remains found in Gandhara, Kucha, Niya, and other regions (Y. Zhang 2006). However, the small domed stupas and arched shrines at the four corners of the pedestal of the stone stupa on a vajra base miniaturize the elements seen in the 5th–6th-century votive stupa in Khocho (The Museum of Indian Art, Berlin, MIK III 610, MIK III 6838; see O’Neill and Laing 1982) and the round stone pagodas in the Northern Liang in Dunhuang (Dunhuang Museum 2017).

The stele-shaped shrine emerged from the fusion of the Gandharan stone Buddhist shrine and the Chinese monumental stone-carving tradition. It was commonly referred to as “statue stele”. There were two types: the single-sided stele-shaped shrine, structured around a Buddha image with a backscreen, and the multi-sided stele-shaped shrine, structured around a stele. The stele-shaped shrine collectively donated by Buddhists in Taiping County, Shanxi, in 549 belonged to the latter (Figure 4). It imitated the stele halo (three curved lines), stele head, stele body, and stele base (lost). Apart from the scene of “Mañjuśrī visiting Vimalakīrti and inquiring about his illness” (Huang 2011) depicted in the wooden, house-style shrine at the very top of the front side, the main decorative elements are symmetrically reproduced on the back and sides of the stele. It embodies three techniques fundamental to Chinese stone carving: relief carving, intaglio carving, and engraving lines. The figures exemplified the slender and elegant style of the Southern Dynasties, making this stele a masterpiece of integrated artistic traditions.

With variations in elements such as pedestals, backscreens, shrine structure, bodhi trees, and curtains, pedestal-style shrines exhibit many variants. For example, two bronze statues dated to the second half of the 6th century unearthed in Boxing County, Shandong, used bodhi trees to construct ritual sites of Buddha. The smaller statue employed symmetrical leaf-like ornaments to frame three Buddha images, while ribbons along the edge of the mandorla formed a shrine-like outline. The larger statue featured a carved openwork bodhi tree, whose intersecting canopies and streams of nectar-like elements framed the shrine (S. Zhang 2009). A similar approach involved single-screen, double-screen, and triple-screen variations. Most artifacts unearthed from Xiude Temple in Quyang, Hebei, were single-screen shrines (The Palace Museum 2009), while gilded bronze shrines in the Northern Wei and Tang dynasties in the Asian Art Museum of San Francisco (accession numbers B60B643, B60B1035, and B60B8+) and the gilded bronze shrine in the Northern Wei Dynasty in the Musée Guimet in France (accession no. EO2604) were multi-screen shrines.

2.2. Hanging-Style Shrine

Without a pedestal, a hanging-style shrine was made to be hung on walls or shelves or carried attached to the shoulder or chest. Its forms included painted silk banners, painted wooden panel paintings, wooden carved panels, and wooden wall shrines. Previously, scholars have primarily focused on wooden carved wall shrines. For example, the Seated Buddha (MIK III 4722) collected by Albert von Le Coq in Khocho and now housed in the Museum of Asian Art in Berlin, as well as two miniature lotus-petal-shaped shrines (Le Coq 1913, Pl. 57: f, e), resemble the Maitreya statue discovered in Bachu County [Figure 5c]. The wooden fragment collected by Stein from the Buddhist temple sites in Yanqi [Figure 5a] retains a bronze nail on its right edge, and traces of nails are also visible on the back, which were used for fixing it to the wall (Stein 1921a, pp. 1202–3). Another wooden fragment Stein collected at Domoko [Figure 5b] features an arched shrine at the top and a curtain-shaped shrine at the bottom, with musical instruments carved on the lintel. The wooden panel is only 2.22 cm thick (Stein 1928, pp. 1053–54) and is likely the upper left end of a wall shrine, serving the same spatial decorative function as the wooden panel paintings unearthed at Domoko.

Small, wearable bronze shrines, which embody a portable ritual site in the minimalist mode, have received less scholarly attention. On the edges of the bronze seated Buddha collected by Albert von Le Coq in Khocho, there are four equally spaced small holes, allowing it to be sewn onto clothing, headgear, or ritual objects with a needle and thread [Figure 5d]. The small bronze seated Buddha discovered by Stein in Yotkan [Figure 5e] measures only 2.5 cm in height and retains a clasp-like component on its back (Stein 1921a, p. 103), indicating that it is a wearable object.

Silk banners were mainly found in Dunhuang and Turfan.1 Previously, scholars have largely regarded them as independent paintings and failed to recognize their ritual functions. In fact, banners were used in military and sacrificial ceremonies2 or as the sign of Buddhist ritual sites. For instance, the Prince of Zhongshan in the Western Jin Dynasty led monks in “welcoming sutras with silk banners on the outskirts of the city” (Yan 1958). In the Tang Dynasty, the Han temple in Yotkan employed artists to decorate “processional ritual banners and canopies”.3 Zhi Dun 支遁 (314–366), an eminent monk, used the phrase “banners descending at the appointed time” (F. Zhang 2014) to emphasize that Maitreya inherited the orthodox Dharma. Thus, silk banners also constituted an important motif in cave carvings and murals across various regions. According to artifacts, a silk banner generally consists of four parts: (A) Banner head: Slightly wider than the banner body, edged with brocade, often featuring a Ren (人)-shaped sloping roof of a shrine, with a triangular lintel. When the banner body is wide and difficult to shape, only 3–5 clasps are used for hanging. (B) Banner body: Primarily single or assembled silk paintings, edged on both sides, sometimes with additional ribbons on the outer edges. Narrower banners often feature standing images of the Buddha and Bodhisattva, with musicians and donors painted in the corners [Figure 5f]. Wider banners depict the scenes of Buddha preaching [Figure 5g]. (C) The bottom of the banner: The hanging sections are attached to the lower part of the banner body, with a weighted board at the very bottom to maintain shape or prevent curling. Complete examples include Ch.002, Ch.004, Ch.liv.002, etc. (Stein 1921a, Pl. LXXXII, LXXXVI.) (D) Banner: A rod used to connect the banner head or body for fixing, hanging, and moving. Wider painted banners fixed to walls or wooden boards often omit this component. Stein keenly observes that these meticulously crafted banners, while clearly representative of Han Buddhist artistic conventions, suffer from a notable homogeneity of form and a lack of distinctive individual expression (Stein 1921a, pp. 937–38). This is partly because they share similar preparatory sketches with cave murals and statues, displaying traits of mass production (Jao 1978; Sha 2007). On the other hand, as offerings made by Buddhists, they emphasize devotional rituals or periodic exorcism ceremonies rather than serving as objects for scholarly appreciation (Eiichi 2019, p. 3).

2.3. Open–Close- or Mother–Child-Style Shrine

The mother-and-child style is mainly formed from or based on niches, reliquary boxes, lotuses, etc. After the central part is hollowed out, a Buddha statue is placed inside, or small boxes are nested layer by layer to accommodate the child components. Although the mother and child components can be separated, they maintain a complete compositional relationship. Examples include the schist stone carvings unearthed from Pakistan and Kashmir, dating to the 5th–7th centuries, housed in the Metropolitan Museum of Art (accession no. 1987.142.51, 1987.142.53, 1987.142.66, 1994.489, etc., see Behrent 2007), as well as an ivory shrine unearthed in Anxi, Gansu [Figure 6a]. The Tang-dynasty stone shrine [Figure 6b] in the collection of the Asian Art Museum features a central hollow structure designed to hold a miniature Buddha image. More iconic examples are the eight-layered and five-layered precious caskets unearthed from the underground palace of Famen Temple Pagoda in Chang’an (Han 2012), which attest to the tradition of container-like shrines for housing Buddhist relics. Slightly later examples include the lacquer-carved shrine and the wooden lotus-shaped shrine from Chongsheng Temple in Dali, Yunnan [Figure 6c]. The former contains a gold statue of Avalokiteśvara, while the latter is carved with five layers of lotus petals and two arches, housing three miniature Buddha images inside (Jiang and Qiu 1998).

In addition to the above-mentioned types of shrines, plaster and clay eave tiles, as well as “virtue mud” 善业泥 tablets, could be transformed into portable shrines. These were often painted by devotees and enshrined for worship. There are also examples where shrine elements were incorporated into everyday objects, such as inkstones and stone foundations unearthed in Datong. These can be regarded as derivatives of independent shrines.

3. Multi-Modal Visual Grammar Shaped by Local Traditions

Regarding portable shrines from the 3rd to the 8th centuries along the Silk Road and neighboring regions, only a few artifacts have been discovered at temple sites, while the origins of many examples remain unclear. Not only have these objects from hoards or burials been disconnected from their primary contexts, but their placement in museum displays also renders the task of tracing their original functional and formal relationships significantly more difficult. Some objects carry valuable information about their donors or users. Such evidence facilitates a visual–anthropological analysis of their production, use, and cultural significance.

Given that design and decoration are central to object-creating, our examination will commence with an analysis of their visual grammar. While the concept of visual grammar—adapted from linguistics and semiotics—remains insufficient to capture the full complexity of portable shrines, it nonetheless provides a valuable framework for analyzing the cultural conventions that govern their artistic expression. According to Roland Barthes, visual grammar participates in a system of signs built from images, gestures, musical sounds, and objects. This system shapes the content of ritual, convention, and public entertainment, acting as a kind of “isology” that carries meaning much as language does (Barthes 1968, pp. 9, 43). This terminology further accounts for experiential and individual differences, while highlighting its role as a medium for communication and social interaction (Hatcher 1999; Kress and van Leeuwen 2021). This corresponds to the multi-modal grammatical framework we intend to explore, a framework that is both grounded in and extends beyond the visual dimension.

3.1. The Interplay Between Images and Inscriptions: From Static to Dynamic

The early portable Buddhist shrines along the Silk Road and neighboring regions drew upon a creative synthesis of local traditions, specifically by integrating object-centric ritual practices with commemorative art. Hava Katz’s study reveals that portable clay shrines from the Near East dating to before the mid-1st century already displayed three main forms—closed, Noas,4 and open models—each with several subtypes (Katz 2016). This suggests a direct link to the emergence of Buddhist art. The images of demigods, dwarfs, and rats on Buddhist architectures and statues unearthed at Amravati, India (Knox 1992, p. 42); the blending of Gandharan and Central Asian shrine models at Buddhist sites in the Swat Valley of Pakistan; and the representation of Indian deities on portable stone steles (Brancaccio 2018, pp. 202–3) all demonstrate how regional object-creating traditions supplied essential schemas, motifs, techniques, and even the impetus for the evolution of Buddhist art. Swāti further delineates four distinct traditions in the Buddhist statues in the Swat Valley: originating with the Hīnayāna and Mahāyāna (the cults of the Buddha and Bodhisattvas), turning to the Tantrayāna and the Vajrayāna centuries later. He then classifies the style of Gandharan Buddhist art, referring to the Indus-Oxus School, regional styles, and many local workshops (Swāti 1997). Simultaneously, Buddhist carvings and murals unearthed in Khotan feature images from Gandhara, such as Indra, Hariti, and the rat king, alongside animal motifs like wolf, sheep, and lions, commonly seen across the Eurasian steppes, incorporating elements of Zoroastrianism and other indigenous religions (Bureau of Cultural Relics of Xinjiang Uygur Autonomous Region 2011). This phenomenon mirrors that in 2nd–3rd-century China, Buddhism was initially equated with supernatural forces, and Buddha images gradually began to supplant icons of traditional deities (Wu 1986).

Transforming a deity shrine into a Buddhist one demanded not only a shift in the pantheon but also an adaptation of its iconography. However, the cognitive framework—encompassing the function, schema, and symbolic sacredness of the “shrine”—remained fundamentally unchanged. In his study of the Han dynasty cliff tombs in Pengshan County, Sichuan, Liang Sicheng described the hollowed-out miniature stone chambers as functionally resembling “small niches or hearths” (Liang 2011). In Zhongxian County, Chongqing, clay houses with Buddhist elements dated to the 3rd century have been discovered (Sichuan Provincial Commission for the Preservation of Cultural Relics 1985). Similar shrine-like structures were also assembled using stone reliefs, as seen on the back wall of the front chamber of the Wu Family Shrines in Jiaxiang, Shandong, dating to the mid-2nd century. The offering table beneath the small niche and the images on its walls mediated between divine and ancestral spheres (Jiang and Wu 2000). Whether a shrine was carved or painted, it primarily offered a conceptual framework. Its ritual significance and content, however, were defined by communities through the use of particular images and inscriptions.

The practice of inscribing donor identities and vows directly onto Buddhist shrines originated in India and Central Asia5 but flourished at the eastern end of the Silk Road and neighboring regions. Evolving from earlier traditions of familial shrine architecture, stone carving, and ornamental craft, portable Buddhist shrines with “statue inscriptions” first appeared in present-day Henan, Shaanxi, Shandong, Hebei, and Shanxi in the 4th and 5th centuries. The practice spread widely throughout China and became institutionalized as a standard form of collective offering by kinship or local groups, often featuring engraved prayers and the names of donors. With fewer donors, as seen in the shrines dedicated by Zhiming 智明 and Wuqiu Haishen 毋丘海深 (Figure 7), the design of the pedestal could be significantly simplified. However, when donor groups were large, the inscriptions—as seen on the steles donated by Taiping County (Figure 4), Yong Zizhi 雍子直 (Figure 8), and Wei Wenlang 魏文朗 (Figure 9)—expanded to fill more decorative space, becoming an integral part of the visual design.

Inscriptions materialized faiths and rituals, creating a convergence where the donor’s worldly aspirations met religious sacredness, as evidenced in three aspects. First, inscriptions could simultaneously serve multiple purposes: commemorate deceased family members, offer blessings for emperors, and pray for the welfare of the clan and all sentient beings. For example, Wuqiu Haishen began by wishing “peace and well-being for the family” (願家口平安), followed by prayers for Wu Zetian, “for the Celestial Emperor and Empress, may all lands be governed” (上為天皇天后，𣏌楃萬邦), and then for “deceased parents,” “souls of seven generations,” and “the entire household.” The inscription also follows the official document writing format, starting a new line at “Heavenly Emperor and Heavenly Empress” (sixth line). Similarly, the votive shrine donated by Yong Zizhi and others also included prayers “for the Emperor first, then for monastic teachers and parents, all sentient beings, that they may together attain perfect enlightenment” (上為皇帝陛下，又為師僧父母，一切眾生，普同正覺), with a new line beginning at the characters for “Emperor” (Figure 8, top right). Second, a visual hierarchy was observed in the listing of donors’ names, which adhered to contemporary social rankings. Arrangement followed either official ranks or community roles—including community leader, community preceptor, and community members—directly reflecting distinctions in status. To ensure no donor was omitted, two approaches were viable: extending the list onto the stele’s edges, or—following the model of the votive shrine donated by Yong Zizhi—prioritizing the names of the donors rather than the coherence of the prayers. Third, the diversity of donors’ beliefs bred a hybrid style in inscriptions and visual design. On the front side of the stele donated by Wei Wenlang, Buddhist and Daoist deities were enthroned side by side (Figure 9). Wei Wenlang identified himself as a “devout Buddhist disciple”, yet his household included Daoists (Shaanxi Yoxian County Yaowangshan Museum et al. 1996). To accommodate the devotional needs of all family members, the stele necessarily incorporated deities drawn from Buddhist, Daoist, and popular religious traditions. Likewise, the stone votive shrine unearthed from a tomb in Datong [Figure 3c] was a product of a diverse belief system: a shrine that combined Xianbei6 military exorcism rituals and folk burial customs (Wang and Cao 2004; Cao and Wei 2022). This connected Buddhist art in Pingcheng in the Northern Wei Dynasty and shamanistic art.

More importantly, although inscriptions and images were presented in a static form, they were generated dynamically. First, after Buddhist shrines were constructed, they were handed down among other devotees, who later added decorations or inscriptions. The original appearance of the shrine donated by nun Zhiming cannot be determined, as it may have been painted later. The closing line, “While Guo Da and Guo Hu serve Buddha” (郭達郭胡侍佛時), indicated that donor information was added afterward by someone familiar with them [Figure 7a]. In contrast, the votive shrine donated by Yong Zizhi was handed down among devotees for over one thousand years. In the first lunar month of the 14th year of Emperor Zhengde of Ming (1519), villagers such as Hou and Liu from Wanxian County, Baoding Prefecture, Zhili, added the inscription “made on an auspicious day” (吉日造) on the left side of the pedestal (Figure 8, middle right). Second, from its initial completion, through each reconstruction as a portable ritual site, to subsequent decorations, the shrine was embedded within a network of devotional acts, including Buddha veneration, vow-making, and disaster-averting ceremonies. Most ritual activities generated texts such as community feast records, votive texts, and merit records (Ning and Hao 1997). Some large-scale public donations mobilized devotees from multiple neighboring temples and villages, in which various activities were staged, such as erecting steles, repairing pagodas, chanting sutras, worshipping Buddha, and planting pines.7 These events involved not only texts and inscriptions separate from the shrines but, more crucially, rituals like recitation and chanting. Such performative practices, transcending the merely visual, continually reshaped the meaning embodied in the shrine’s images and inscriptions.

3.2. Complementarity and Imitation in Visual Form

Portable Buddhist shrines inevitably evolved toward stylistic similarity once their design was standardized and their techniques were conventionalized. Aside from a minor difference in the height of Buddha images, the two small white clay shrines discovered in Boxing, Shandong [Figure 2d,e], are otherwise identical in form. This indicates that their molds were derived from a common prototype. Similar shrine forms can fulfill basic devotional needs but cannot satisfy donors’ aspirations for exceptional craftsmanship and the worldly prestige it signified. Hence, artisans strove for a delicate synthesis—comprehensive in content yet personalized in form—in their effort to realize the following ideal: “Extraordinary in appearance and complete with the thirty-two marks of Buddha, Bodhisattvas stand solemnly in attendance, and heavenly beings soar, filling the scene. Beholders lose themselves in contemplation as divine light shifts westward, while observers admire the interplay of dawn light and shadow.”8

The simplest and most effective method was to apply painting and gilding to Buddhist shrines, thereby transforming an ordinary object into one that seemed “radiant with a descending divine presence 毫光載臨” (Dong 1983, p. 4370). Some white-clay Buddha images unearthed in Boxing, Shandong (e.g., samples 753-280, 341-162, 763-290, 487-58, 483-54, 484-55, etc.), retain traces of red pigment applied after formation (Zhang and Xiao 2011). The figures and base plates of the hanging-style wooden shrines mentioned earlier [Figure 5a,b] were also enhanced with red, black, or blue pigments (Stein 1921a, pp. 1202–203). Wooden panels (e.g., Har.050) and small clay shrines (e.g., Ch.01, Ch.020–022, Ch.025, and Ch.031) discovered by Stein in Dunhuang, Khotan, and other regions (Stein 1928, pp. 361–62, 1055–56) also showed evidence of gilding and painting.

The painted decoration on portable shrines adeptly employed two techniques: color filling and structural painting, both of which complemented the sculptural grammar. The former applied color to sculpted surfaces. This approach was mainly seen in the painted white-clay and wooden examples mentioned earlier, the stupa-shaped shrine unearthed from Dunhuang [Figure 2a], and the votive shrine donated by Yong Zizhi and others (Figure 8). The latter used painting to enhance or enlarge the shrine’s structure or to add other image motifs, which became an integral part of the shrine. The White Marble Figure of Bodhisattva (Xin 42929) unearthed from Xiude Temple in Hebei features the donor painted on the backscreen (The Palace Museum 2009, pp. 128–29). The votive shrine donated by Zhiming was also later used with white and red lines to outline the halos, body halos, and sleeve patterns of Buddha and Bodhisattva, achieving greater clarity and depth. Additionally, four monks and nuns facing Buddha with hands clasped in worship were painted on both sides of the shrine [Figure 7a]. This not only reinforced the presence of donors but also re-defined the spatial layout of the single-screen shrine. Similarly, the votive shrine donated by Wang Zudeng [Figure 10a] employed red and brown lines to delineate the details of the Buddha image, while using black and brown lines to paint a group portrait of donors on the back and left sides of the pedestal. The figures in the painting wore Xianbei-style square hats9 depicting both robed nobles and short-clothed attendants. When connected, they formed a horizontal-scroll figure painting, whose style closely resembled the murals from the tomb of Xu Xianxiu in Taiyuan, Shanxi, dated some 20 years later (Xu 2011). A more ingenious technique involved adding small paintings to the surfaces of Buddha images. For example, on the Northern-Qi Locana Buddha statue unearthed from Longxing Temple in Qingzhou, small seated Buddha images, flying apsaras, and scenes from Buddha’s life were painted within the gilded grid patterns of the robe (Liu 2000). A slightly smaller Locana statue [Figure 10b] also employed this method. In fact, on the statue of Locana unearthed from the Balawaste Buddhist Temple in Khotan, the painter depicted auspicious symbols using lines—such as the wish-fulfilling jewel, vajra, stupa, and galloping horse—to create an illusion of three-dimensionality in the mural figures [Figure 10c]. This approach aligned with the original meaning of the scripture: “All Buddhas of the ten directions and three times manifest their forms within the body of Buddha 十方三世諸如來，於佛身中現色像” (Tripiṭaka Master Buddhabhadra 1985, p. 15). The grammar of extending image structures through symbols may have served as a precursor to the evolution of the Locana Dharmadhātu in Kucha, the Hexi Corridor, and the Central Plains (Li 2000).

The grammar of the imitation of visual form is also evident in painted silk banners. Ch.002 [Figure 5f, (Stein 1921a, p. 938)], which Stein regarded as a standard example, constructed a shrine-like structure at the banner head and simulated a Sumeru stele base using the trapezoidal pendant panel of the banner’s lower section and plant pattern. The Bodhisattva in the painting stands on a lotus, crowned by a canopy, creating a mobile ritual site. A space for inscription was left in front of the Bodhisattva’s face, which was a common technique in Han dynasty stone carving. The large banner Ch.liii.001 [Figure 5g], which depicted the preaching of Buddha beneath the Bodhi tree, did not adopt the outline of the stele-shaped shrine. Stein noted, however, that the halos of Buddha and Bodhisattva were outlined in gold, and their skin was painted yellow and shaded with red. This technique, now faded, creates an impression reminiscent of a bronze statue. More importantly, the Sumeru base of Buddha was obscured by a stele-like object supported by a turtle, featuring an arched top ornament and a slab-stone inscription space (Stein 1921a, pp. 1055–56). Clearly, a stele-shaped shrine was embedded within the painting, which could serve as an inscription space for Buddha or donors, or the pedestal for the entire hanging-style shrine. The form and function of the “stele”10 ensured the continuity of the ritual site of Buddha in Chinese cultural tradition and secular life.

4. Conclusions

The term “historical grammar” was coined by Riegl to characterize the parallel development of human art, polytheism, monotheism, and the scientific worldview: first by enhancing nature through physical beauty, then through intellectual beauty, and finally by emulating and recreating it (Riegl 1966, pp. 13–130, 261–90). At the macro level, Buddhist art corroborates this historical grammar. At the micro level, we observe a complex intertwining of physical, intellectual, and re-creative beauty, as well as a practical reconciliation of polytheistic and monotheistic beliefs. Portable Buddhist shrines from the 3rd to the 8th centuries along the Silk Road and neighboring regions embody the fusion of cultures, regions, and faiths. Itinerant artisans and painters disseminated foreign elements and international styles, while the practical impetus came from donors: royal families, nobles, intellectuals, and rural devotees. By absorbing and blending object-creating notions and artistic traditions, they forged distinct regional Buddhist art styles in areas such as Kucha, Liangzhou, Pingcheng, Yecheng 邺城 (Today’s Handan, Hebei Province), Luoyang, and Chang’an. This evolution culminated in the creative Sinicization of Buddhist art (Jin 2002, pp. 18–34). With the maturation of the Sinicized style, a revival of classical Indian statue styles emerged (Okada 2002, p. 121). Simultaneously, the styles of Luoyang, Yecheng, and Chang’an were disseminated, which, as reflected in portable Buddhist shrines, synthesized a diverse range of concepts, motifs, and styles.

Portable Buddhist shrines provide a critical lens for examining how Buddhist art developed in specific socio-cultural contexts, moving from the analysis of regional styles or local knowledge (Geertz 1983, pp. 94–120) to the reflection of international style. This approach offers a scope for further scholarly discussion. This paper focuses on two fundamental aspects. First, previously overlooked items such as panel paintings, banners, and badges, which, like typical forms such as niches, stupa niches, and stele niches, serve the same or complementary ritual functions in different material and spatial forms and represent an important type of localized development of portable Buddhist shrines, reflecting the interactive relationship between small mobile worship spaces and large cave temples. Second, the production and use of portable Buddhist shrines along the Silk Road and its surroundings integrated early local artistic traditions, forming visual grammar conventions that combined images and text dynamically, which were complemented and mimicked by sculpture and painting. This provides a way for us to understand Buddhist artistic styles from the perspective of visual culture. In fact, the overall interaction of materiality and function and the dynamic complementarity of visual form and grammar may also be a common characteristic of other types of movable religious objects, facilitating diverse connections in cross-regional and cross-religious creation.