Dunhuang Architectural Studies, 1926–2024

Abstract

This field statement reviews the sources, issues, approaches, and missions of Dunhuang architectural studies from 1926 to the present. The studies of Dunhuang architecture constitute a subfield at the intersection of Dunhuang studies, Silk Road archaeology, and studies of Chinese architectural history. Dunhuang architectural studies primarily investigate three corpses of materials, i.e., the images of architecture represented in mural paintings, the cave typology, and the timber-framed façade screening the caves. This study outlines the three phases in which scholarly concerns evolved from images to their material carriers, from forms to spaces, and from disciplinary to interdisciplinary. The first phase, 1926–1950, features Chinese, Japanese, and German scholars’ early explorations of the images of architecture, mostly based on photographic reproductions of murals and timber façades. They established connections between the Dunhuang materials and the modern studies of art and architectural histories. In the second phase, 1950–2000, scholars, mainly Chinese, explored the majority of the primary materials and managed to establish a research framework for research that is still valid today. Some others conducted refreshing studies from the perspective of spatial conception. The third phase, from 2001 to the present, witnesses the emergence of new materials, perspectives, and technologies, which stimulate interdisciplinary and innovative studies of the Dunhuang materials. Finally, this review reflects on the received conceptions the field has brought us to consider the architecture of a Dunhuang cave temple and points out issues of space for future investigations.

1. Introduction

1.1. Scope: Pictorial and Actual Buildings from the Dunhuang Caves

The studies of Dunhuang architecture constitute a subfield at the intersection of Dunhuang studies, Silk Road archaeology, and studies of Chinese architectural history. The term Dunhuang architecture (Dunhuang jianzhu 敦煌建築) was coined by Liang Sicheng 梁思成 (1901–1972), a pioneering scholar of Chinese architecture, in his discussion of the timber-structured porches at the Mogao caves in 1932 [1]. Concurrently, Dunhuang mural paintings that contain images of architecture were studied as an important visual source of nonextant early Chinese architecture and, by the mid-twentieth century, became an independent object of study [2,3]. The full connotation of the term was eventually stabilized when Xiao Mo 蕭默 (1938–2013), a Chinese architectural historian who was a disciple of Liang’s, published a monograph titled A Study of Dunhuang Architecture (Dunhuang jianzhu yanjiu 敦煌建築研究) in 1989 [4]. Dunhuang mainly refers to the Dunhuang caves, as historical architectural remnants in the Dunhuang-Anxi 安西 region are concentrated at the cave sites. They comprise the Mogao caves (Mogao ku 莫高窟, Figure 1) and a few smaller cave sites, including the Yulin caves (Yulin ku 榆林窟), the West Thousand-Buddha caves (Xi qianfo dong 西千佛洞), and the East Thousand-Buddha caves (Dong qianfo dong 東千佛洞) (Figure 2) [5]. The term architecture, while including standalone buildings, mainly denotes the visual and material contents of the caves that represent architectural elements. Thus, the composite term Dunhuang architecture incorporates two main categories of materials: (1) the images of architecture (jianzhu tuxiang 建築圖像) represented in mural paintings (Figure 3) and (2) the ancient architecture in Dunhuang (Dunhuang gu jianzhu 敦煌古建築) including the cave temples (Figure 4), their timber-framed porches and ante-halls (Figure 5), and ancient pagodas (Figure 6) [6] (pp. 47–55). The visual and archaeological sources provided by the Dunhuang caves were produced in the same historical space, but they have often been studied as separate categories.

1.2. Historical Accessibility and Assessments

The limited accessibility of the primary source is an apparent historical reason for the discrete studies of the two types of sources. Although Dunhuang is believed to have been a prosperous crossroad of the ancient Silk Road [7], it has been generally accessible to modern visitors only in the last decades of the twentieth century, when railways and airlines began to connect it to major cities [11]. Thus, in the first half of the twentieth century, those who explored the cave sites were not specialists in architectural studies, whereas those who investigated the architecture-related sources from Dunhuang did not have a chance to explore the cave sites in person. The visual sources made available to early scholars of Chinese architectural history were mainly black-and-white photos of the Mogao caves taken by the early explorers—including archaeologists and Sinologists—for their own interests, which did not necessarily include architecture. The most cited source was the six-volume photo album titled Les grottes de Touen-houang: Peintures et sculptures Boudhiques des epoques des Wei, des Tang et des Song and published in Paris in 1914–1924 by French Sinologist Paul Pelliot (1878–1945), who stayed at Mogao for three months in 1908 [12]. In comparison, the visual documentation of the cave spaces was not exploited as efficiently as possible. The reports of the two most thorough archaeological surveys of the Mogao cave complex—one conducted by Russian archaeologist Sergei Feodorovitch Oldenburg’s (1863–1934) expedition team in 1914–1915, the other by Chinese archaeologist Shi Zhangru (Shih Chang-ju) 石璋如 (1902–2004) in 1942—were not published until the 1990s [13,14]. The value of these two reports for current studies of Dunhuang cave architecture cannot be overstated and will be discussed later in this review. Nonetheless, in the first half of the twentieth century, the lack of concrete information on the physical forms of the caves or any spatial experience of the cave site hindered early scholars from studying the cave architecture of Dunhuang.

In comparison to the actual architecture, the images of architecture have received more scholarly attention for not just easier accessibility but also for unparalleled richness. About a third of the decorated caves bear mural paintings that contain architectural elements, and as many as three to four hundred transformation tableaux contain pictorial depths defined by the images of architecture. The Dunhuang images of architecture encompass a variety of building types, such as palaces and monasteries, gate towers, walled cities, pagodas and stupas, residential buildings, prisons, tombs, cottages, and bridges. Because of the encyclopedic scope of the contents, the Dunhuang images of architecture offer points of investigation for the layouts of architectural complexes, the nonextant building types, the architectonic details, the construction process, and the precursors of architectural painting, historically known as ruler-lined painting (jiehua 界畫).

The persistent interest in the Dunhuang images of architecture is grounded on the scarcity of extant early timber-framed buildings in China. Chinese architecture is characterized by the timber-framed construction system and the arrangement of individual buildings into courtyard complexes, but only very few timber-framed buildings from the first millennium CE are extant [15]. As it is believed that Chinese timber-framed architecture “matured into full glory and vigor” in the Tang dynasty (618–907), the architectural developments of that golden period and before were of primary interest to early studies of Chinese architecture [16]. However, only a handful of Tang-period timber-framed halls are extant, and no building complexes from the Tang dynasty or before still stand in their original appearance. According to a recent survey, no more than nineteen timber-framed buildings dated before the end of the tenth century survive [17]. Thus, the foremost and best-studied subject matter among all types of images of architecture is the images of palatial complexes in the scenes of Pure Lands (jingtu 淨土) or Buddhist paradises [18]. An irreplaceable historical value has been bestowed on the visual representation of architecture ([2], p. 75, [6], p. 60). As Liang unreservedly praises them, the Dunhuang architectural images are “the second best, most faithful, and most valuable materials that are only inferior to actual buildings (……次於實物的最好的、最忠實的、最可貴的資料)” [3] (p. 2). In this phrase, “actual buildings” (shiwu 實物) refer to freestanding architecture in China in general, as opposed to the archaeological materials of architecture in the Dunhuang Caves.

1.3. Overview of the Field’s Developments and Inquiries

While studies of Dunhuang architecture have advanced in the past ten decades, questions of how to understand the correlation among the materials have gradually surfaced. In the early phase, the limited availability of Dunhuang visual materials confined the architectural studies to the image of architecture. Subsequently, when the source problem was alleviated, the cave’s architecture still could not secure a decent position in the architecture family, as it was studied either as the imitation of stand-alone architecture or the embodiment of a concept of space fundamentally different from that of architecture. Thus, the image of architecture and the cave architecture became separate categories in a taxonomy oriented to the topological studies of stand-alone architecture. Studies in the current century have challenged the boundary by comparing and combining architectural information extracted from the two categories of materials. Concurrently, the abstraction of compositional principles hindered the perception of the total space a cave possesses. To anchor the architectural studies of the Dunhuang caves to historical realities, clues might be drawn from the paralleling methodological explorations in monastery studies, cave archaeology, Dunhuang studies, architectural conservation, art history, and digital humanities.

The following three sections will unpack the subfield’s developments in three successive phases. The last section will reflect on the missions, achievements, and limitations of the current scholarship, shedding light on future directions.

2. The First Phase (1926–1950): Early Exploration of Materials and Methods

2.1. SRCA Scholarship Shaping the Discourse About Chinese Architectural History

Owing to the relatively high publicity of Pelliot’s photo album, pioneering scholars in Japan, China, and Germany began to incorporate the new materials from Dunhuang into their studies of timber-framed architecture as early as the 1920s–1930s. Western-trained scholars based in East Asia were sensitive to the new materials and competed to extract information about ancient architecture from them. Hamada Kōsaku 浜田耕作 (1881–1938), a Japanese archaeologist trained in England, was one of the first scholars to have consulted the Pelliot photos for architectural studies. In a 1926 article, Hamada brought in Dunhuang porches and murals of architecture as continental references for the early timber-framed architecture of Hōryūji 法隆寺 Temple in Nara, Japan [19].

Stimulated by European and Japanese pioneers’ scholarship, research on ancient architecture in China by native scholars accelerated in the 1930s. A crucial step was the establishment in 1931 of the Society for Research in Chinese Architecture (Zhongguo Yingzao Xueshe 中國營造學社, hereafter SRCA), China’s first academic group specializing in traditional architecture [20]. The scholar-members of SRCA not only conducted studies of architectural treatises and historical buildings but also actively introduced fresh studies from overseas. Most of their work was published in the SRCA’s bulletin, Zhongguo yingzao xueshe huikan 中國營造學社彙刊 [Bulletin of the Society for Research in Chinese Architecture], from 1930 to 1945 [21]. Hamada’s article was translated into Chinese and commented on by Liu Dunzhen 劉敦楨 (1897–1968), a scholar of Chinese architecture trained in Japan and a founding member of SRCA [22]. Liu’s substantive commentary indicates a mixed attitude of recognition and emulation from the SRCA side. It is exactly for the reason of comparison and competition that the first article examining Dunhuang architecture, written by a Chinese author, was published alongside that Chinese translation of and commentary on Hamada’s article. Both were published in the SRCA’s bulletin in 1932. That article, written by Liang Sicheng, a US-trained scholar and another founding member of SRCA, aimed to reveal the general features of Buddhist monasteries and palaces of Tang China [2].

Despite not being the earliest scholarship to utilize the Dunhuang materials, Liang’s article was the first to investigate the full-scale historical value of the materials. While Hamada focused on the style evolution of bracket sets represented by Dunhuang mural paintings [22] (p. 9), Liang used them, together with other textual and visual evidence, to demonstrate the features of the Tang Dynasty architectural system. This system contains three scales (from large to small): the layout of building complexes, the prototype of individual buildings, and the architectonic details (Figure 7) [2]. Both Hamada and Liang cited the cave porches at Mogao as a reference for structure design before the advent of architectural treatises in China, yet Liang further attributed them to the Tang style by consulting Pelliot about the dated inscriptions ([1], [22], p. 27). It is the system based on scale and typology and the great interest in Tang-style architecture that has had such a profound impact on subsequent studies of Dunhuang architecture.

2.2. An Art Historical Glimpse into Religious Visual Culture

Taking Pelliot’s Dunhuang album and Liang’s first article as references, German-American art historian Anneliese Bulling (1900–2004) published a dissertation in 1936 on early Chinese architecture [23]. Bulling’s work represents the first effort made by a non-Chinese scholar of Chinese art and architecture to systematically examine the materials of Dunhuang architecture. The dissertation, titled “Die chinesische Architektur von der Han-Zeit bis zum Ende der T’ang-Zeit”, analyzed the formal features of Chinese architecture in more than one hundred images of architecture, some fifty of which were from Dunhuang. Bulling expanded the spectrum of architectural typologies to include city gates, homesteads, and ceiling designs, among others. She also offered more variants for each typology that Liang had mentioned. However, there are discrepancies between the formal analysis and the scope of the dissertation, which was to explain the stylistic development and metaphysical characters of Chinese architecture from the second century BCE to the ninth century CE. Hence, Bulling’s dissertation has not been widely acknowledged in the field of Chinese architectural history, except for two not-so-positive reviews [24,25]. In comparison, a better-known work of Bulling’s on Dunhuang architecture is a 1955 article titled “Buddhist Temples in T’ang Period”, which adopts a more focused scope based on the core materials in her dissertation [26,27]. Bulling’s 1955 article revisited the issue dealt with in Liang’s 1932 article: what the Dunhuang murals of architecture can tell us about the Buddhist monastic architecture in Tang China. Whereas Liang focused on architecture per se, Bulling addressed the role of Buddhist culture in China in shaping monastic architecture and even Chinese architecture in general. She especially points out the active role of the Buddhist vision of the paradises on earth [26] (pp. 121–122).

Bulling did not just engage with the burgeoning explorations of Dunhuang materials; the European discourses of Chinese Buddhist monasteries also shaped the line of inquiry. Two European scholars who had extensively traveled and worked as architects in China—Ernst Boerschmann (1873–1949) and Johannes Prip-Møller (1889–1943)—urged that close attention be paid to the intersection of architecture and religious cultures.1 Boerschmann and Prip-Møller had engaged deeply in surveying the still active monasteries, sacred mountains, and contemporary Buddhist practices in China, seeking ways to demonstrate the interconnections among architecture, landscape, and religious culture [28].

In the first decades of the twentieth century, German architect Boerschmann studied the contemporary Buddhist cult center Mount Putuo 普陀山 and the ancestral temples in China [29]. As architectural historian Eduard Kögel aptly comments, Boerschmann followed “a holistic approach to cover a still living culture [30]”. For this reason, he did not attempt to find a systematic historical and stylistic order as most scholars of Chinese architecture would undertake in the next decades. This line of thought was echoed by Danish architect Prip-Møller’s Chinese Buddhist Monasteries, which was published in 1937. Prip-Møller presented a picture of “a monastery as an organism living in the present but with its roots deep in the past [31]”. Thus, he chose a monastery that had been substantively rebuilt and renovated between the seventeenth and the twentieth century, but that reflected the architectural layout that could be traced back to the ideal monastery conceptualized by Chinese monks in the seventh century [32]. The common interest of Boerschmann and Prip-Møller lay not so much in the material remnants of ancient buildings as in the built environment through which the way of religious life and the cultural memory of ancient times has been handed down to the present day.

Bulling’s Dunhuang studies, which dealt with art-historical materials rather than architectural and ethnographic surveys, added a new dimension to the holistic studies of architecture and religious culture. Although her discussion is brief and general, Bulling made at least one important contribution: reading the historical mentality—a Buddhist vision of paradises on earth—behind the pictorial and picturesque “monasteries” in the Pure Land scenes and associating it with the pursuit of beauty in actual architecture. The historical dimension is crucial, especially because the lack of early works of architecture has been one of the main reasons for the criticism of Boerschmann’s and Prip-Møller’s scholarship from contemporary art historians. Since their lines of inquiry were far from the mainstream of studies in the twentieth century, their impact on studies of Chinese architecture or Dunhuang architecture has been very limited. Nevertheless, Bulling’s scholarship, especially her dissertation, represents the most systematic exploration of Dunhuang architecture prior to Xiao’s A Study.

Furthermore, Bulling invented a new method to analyze the architectural design of the buildings represented in pictorial images. She made a series of plan and elevation drawings to diagram her conceptual construct of the architectural spaces represented in Dunhuang murals (e.g., Figure 8). Modern architectural drawings of consistent scale and standard viewpoint are meant to translate the pictorial depth in an ancient painting into a three-dimensional space that is imaginary but retrievable. This method, known as plan reconstruction, plan deduction (pingmian tuiyan 平面推演), or spatial reconstruction (kongjian huanyuan 空間還原), is primarily applied to studies of the layouts of building complexes and has been widely utilized in scholarship of the current century ([25], pp. 151–152, [33], pp. 15–16). This method has become a crucial first step to preparing a common ground for extracting visual information about historical architecture, but researchers who use the method often hesitate to address the true implications of manipulation, both its danger and its necessity. Because no two-dimensional painting can equate to a three-dimensional spatial construct, no matter how realistic it seems, the success of its illusion relies on the viewer’s mental construct of the pictorial images of architecture. Therefore, even the most credible reconstruction—for instance, one that refers to abundant actual examples of historical buildings—cannot be free from individual variation and spatial imagination. As an integral part of the unique methodology in the studies of Dunhuang architecture, the visualizing method foretells the irreconcilable paradox between authenticity and subjectivity in processing visual sources.

3. The Second Phase (1950–2000): Taxonomy Versus Spatial Conception

3.1. Archaeological Fieldwork and Knowledge About the Cave-Front Architecture

Subsequent developments in the studies of Dunhuang architecture are grounded in the exploitation of new visual materials by Chinese scholars. Many impressive surveys were conducted by Chinese artists, historians, and archaeologists in the early 1940s when state-funded wartime surveys favored the strategic western regions of China [34]. Moreover, the first modern research institute based at Mogao—the National Research Institute on Dunhuang Art (Guoli Dunhuang yishu yanjiu suo 國立敦煌藝術研究所, hereafter “Dunhuang Art Institute”)—was established in 1944. Two of the most substantial sets of documentation acquired then are the aforementioned survey of Mogao cave forms by Shi Zhangru and the photographic documentation of the Mogao and Yulin caves by journalist James C. M. Lo 羅寄梅 (1902–1987) with his wife, Lucy (b. 1920, maidan name: Liu Xian 劉先), in 1943–1944. However, since the materials were transported to Taiwan and the United States shortly afterward, they were unavailable to most scholars in mainland China until the turn of the twenty-first century [35,36]. Hence, they, as well as several critical studies by the document holders, have not had as great an impact on the studies of Dunhuang architecture as they should. For instance, Shi’s studies of cave forms successfully combined topological and quantitative analyses to determine the grades of caves and identified the late Tang as a critical period of development of the cave architecture of Mogao [37,38], but the innovative method remains little known.

In comparison, fieldwork conducted in the P.R.C. era (1949–) has had a more immediate effect on the research and management of the built environment at Mogao [39]. The Dunhuang Art Institute was reorganized in 1950 to become the Dunhuang Cultural Relics Institute (Dunhuang wenwu yanjiu suo 敦煌文物研究所). The institute hired two architects, Sun Ruxian 孫儒僩 (b. 1925) and Xiao Mo, in 1947 and 1955, respectively, to conserve and study the Mogao caves. Since then, new materials of Dunhuang architecture have been gradually acquired and have stimulated the development of the field [40,41].2 Moreover, in 1951, a survey team from Beijing comprising archaeologist Su Bai 宿白 (1922–2018) and architectural scholars Zhao Zhenzhi 趙正之 (1906–1962), Mo Zongjiang 莫宗江 (1916–1999), and Yu Mingqian 余鳴謙 (1922—2021) examined the Mogao caves to establish guidelines for scientific conservation. The survey report, which was compiled by architectural historian Chen Mingda 陳明達 (1914–1997) and published in 1955, for the first time provided a general view—brief but accurate—of the changing appearance of the Mogao cliff, the periodization of the Dunhuang cave types based on their architectural features, as well as architectural drawings of representative caves (Figure 9) and the Song-period (960–1279) timber-framed porches (Figure 10) [42,43] (pp. 34–104).3 By pointing out that the most magnificent appearance of Mogao occurred in the tenth century, the reporters presented a dynamic view of the growth of the cave complex. Chen, who visited the Mogao caves alone in 1953, investigated and published about two tenth-century earthen pagodas in neighboring areas [42,44] (pp. 105–109). Since they have been repaired or relocated in the following decades [45], Chen’s surveys contain valuable information about their untouched condition in the mid-twentieth century.

Moreover, archaeological excavations in the 1960s–1990s yielded new evidence of porches, halls, and pavilions designed to screen the entrances of the decorated caves, known collectively as “cave-front architecture” (kuqian jianzhu 窟前建築) [46]. Most excavations were conducted by the Dunhuang Institute/Academy right before conservations of the cliff face and cave architecture [47,48,49]. A majority of the discoveries were published in 1985 in an archaeological report compiled by archaeologists Pan Yushan 潘玉閃 and Ma Shichang 馬世長 (Figure 11) [47]. Dunhuang manuscripts about cave distribution and timber-façade construction have been discovered as well [50,51].

Studies of these archaeological and textual materials focus mainly on the technological-stylistic features and the layout. One approach investigates the timber-framed construction system through analyses of extant porches ([4], pp. 269–302, [52,53]), theoretical reconstructions of several non-extant ante-halls [54,55], and reconstructions based on a cave-eave construction memo [51,56]. Another approach surveys the overall distribution of caves on the cliff surface and the historical contexts at play during their construction and use [57,58,59,60]. These studies elucidate at the micro and macro scales, respectively, the splendid appearance of the Mogao caves around the tenth century.

The aforementioned studies indicate that the spatial experience of the Mogao site makes one fully aware of the innovations of the late- and post-Tang caves by their unprecedented scales and magnificent appearance, if not by the visual contents. The time lag between the maturity of the image of architecture (the high Tang, ca. the 8th century) and the peak of cave architecture (the Guiyijun period, 851–1036 CE4) seems curious, yet few comparisons have been attempted. The material forms of caves have often been studied as sources for cave archaeology, whereas the images of architecture have been studied as sources for architectural history unless a way of relating the cave to stand-alone architecture is explored.

3.2. Establishment of a Comprehensive Research Framework

The task of articulating the relationship between cave architecture and stand-alone architecture was first attempted by Xiao Mo, who studied architecture at Tsinghua in 1955–1961, 1978–1981, and 1987–1990 and worked at the Dunhuang Cultural Relics Institute between 1963 and 1978. Xiao, a rare figure who has systematic knowledge of both Chinese architecture and the Dunhuang materials and conducted both extensive fieldwork and academic reflection, brought the studies of Dunhuang architecture to a new level.

For one thing, Xiao established the conceptual framework and methodology for the study of Dunhuang architecture, which are still valid today. In his master’s thesis, Xiao articulated three core objects of study of Dunhuang architecture: cave typology, images of monasteries in Dunhuang murals, and timber-framed porches of the Tang and Song. Xiao’s master’s thesis, written at the beginning of the 1980s, comprised three articles on these three subject matters [61,62,63]. In A Study of Dunhuang Architecture, a revised collection of essays that constituted his master’s thesis and PhD dissertation, Xiao developed his studies of the images of architecture into encyclopedic and typological studies of ancient Chinese architecture (Figure 12). The studies of cave, cave-front, and stand-alone architecture were not expanded as such, but since the publication of the second edition of A Study in 2002, they have been grouped together into “ancient architecture of Dunhuang” as the counterpart of the image of architecture. Furthermore, Xiao established a method to discern the faithful and fanciful aspects of the image of architecture and thereby made the study of the “realistic” aspects of Dunhuang architecture a valid academic inquiry in its own right [6] (pp. 62–63).5 Since Xiao, the approach to studying Dunhuang architectural materials has made a fundamental shift from the early scholarship. Liang and Bulling used Dunhuang materials to prove the general features of Chinese architecture as indicated by historical texts and comparative cases, whereas Xiao used historical texts and comparative cases to prove the features indicated by the Dunhuang materials. The shift of the object of the study suggests the researchers’ increased confidence in the Dunhuang architectural materials and in their capability of extracting valid historical data from them.

Xiao also associated the development of cave types with the evolution of Chinese monasteries. Based on architectural prototypes of the caves, he defined the major cave types in Dunhuang as (a) central-pagoda-pillar cave (zhongxintazhu ku 中心塔柱窟); (b) upturned-funnel cave (fudoushi diantang ku 覆斗式殿堂窟), alternatively known as truncated pyramidal ceiling cave in English literature; (c) backscreened (central-altar) cave (beipinshi ku 背屏式窟); (d) vihara cave (piheluo ku 毗訶羅窟); (e) nirvana cave (niepan ku 涅槃窟); and (f) colossal-Buddha image cave (dafo ku 大佛窟) (Figure 13) ([4], pp. 33–60, [62]). Among them, the first three types are believed to be modeled after the pagoda-centered cloister, the canopied shrine, and the Buddha hall with a backscreened altar, respectively. Because these three types represent the paradigmatic cave designs of the pre-Sui, Sui-Tang, and post-Tang periods, the succession of cave types is treated as indirect evidence of the development of Chinese monasteries. Scholars’ consensus about the latter is that because Buddhist architecture of Indian origin was adapted to Chinese building traditions, a pagoda-centered layout was gradually replaced by an image-hall-centered layout [64] (p. 21). The evolution of cave types that Xiao proposed, especially the genealogy of the backscreened cave, has not satisfied all archaeologists.6 They have proposed other ways of categorization according to the formal elements (e.g., plan, ceiling, niche, altar) of a cave or the functions of caves (e.g., meditation, pagoda circumambulation, image hall) ([14], vol. 1, pp. 4–7, [65,66,67]). But it is Xiao’s categorization that best serves as an analogy between the Sinicization of cave temples and that of Buddhist monasteries.

Xiao built a solid foundation for material-oriented encyclopedic studies of Dunhuang architecture and found a way to describe the architectural features of the cave. Nonetheless, he remained silent about the possible relationship between the image of architecture and the actual architecture of Dunhuang, although both are believed to represent the same types of stand-alone architecture, such as the monastery and the pagoda. An issue that was immediately addressed by contemporary scholarship is the dialectical relationship between artistic creativity and historical reality.

3.3. Visual Representation and Spatial Conceptions

Xiao’s methodology of extracting objective information from works of art was challenged by a PhD dissertation written by architectural historian Puay-Peng Ho at the School of Oriental and African Studies, the University of London, in 1992 [68]. Incorporating the new materials of Dunhuang images of architecture published by Xiao, Ho’s dissertation sought to understand the spatial conception that prevailed in actual Buddhist architecture and the textual and visual representations of them. For Ho, the Dunhuang murals of architecture are not just faithful representations of historic buildings; he also valued them as “an expression of artistic creativity” [68] (p.167). Thus, although most scholars had regarded the palatial complexes in Pure Land scenes as representations of “Buddhist monasteries” [2,26], Ho and those more cautious about the nature of the images called them “architecture in the Pure Land scenes” or “the Pure Land architecture” [68]. Despite the difference in opinion, Ho’s study still benefited from the visual information Xiao extracted from Dunhuang mural paintings. Unlike Bulling’s plan reconstruction, Xiao’s trace-copy line drawing of the Pure Land architecture preserves both the pictorial composition of the mural painting and the architectonic details of the visionary architecture (Figure 14).7 The choice of the latter method implies the notion that knowledge of architectural forms cannot replace the visual effect of architectural imagery. Prior to Ho, the theatrical visuality of Pure Land architecture in Dunhuang murals had let German art historian Lothar Ledderose ponder its potential connection with the heavenly architecture represented in Christian church mosaics of the Roman empire [69]. Ho chose a historically grounded approach to analyze one of the cultural roots of the Dunhuang visual representations, namely, spatial conception.

Ho’s recognition of the value of subjectivity in the visual representation of architecture led to two sets of conclusions. One concerns the notions of spatial conception, including the notions of centrality, duality, verticality, multiplicity, and unity [68] (pp. 167–176). By reinforcing these formal and aesthetic principles with supplementary textual and archaeological evidence, Ho generalized some of them as the spatial conceptions of Buddhist monasteries of Sui and Tang China [68] (pp. 326–357). The other, briefly mentioned, considers the mediating effect of the Dunhuang images of Pure Land architecture. As he concludes, “It is precisely this paradoxical quality of the Dunhuang Pure Land paintings, the paradisiac atmosphere created with the not-so-unfamiliar earthly architectural elements, the remote paradise brought tangible to earth, which mediates and communicates to the viewers the indescribable heavenly bliss [68] (pp. 180–181)”.

The apparatus of such transformative imagery was further articulated in a 1995 article titled “Paradise on Earth” [70]. In this article focusing on Pure Land depictions in the high-Tang Dunhuang murals, Ho argued that the architectural imagery bridges the mundane and supermundane worlds by means of the convincing spatial representation, the notion of duality, and the iconology of magnificence and splendor. Ho’s studies sought ways to emancipate the image of Pure Land architecture from a position subordinate to actual Buddhist monasteries. Bulling had called for recognition of the Pure Land image’s stimulative effect on beauty in the actual built environments [26] (pp. 121–122), and Ho furthered it by specifying the shared concepts of space among the visionary and real buildings.

3.4. Social History and the Production of Cave Space

In the meantime, Dunhuang scholars and art historians began to reformulate the humanistic values of Dunhuang art and to adopt a positive attitude toward the aspirations and visions of the cave makers. Shi Weixiang 史葦湘 (1924–2000), an influential researcher of the Dunhuang Academy (Dunhuang yanjiu yuan 敦煌研究院)—which the Dunhuang Institute was expanded to be in 1984—was one of the main advocates of the social art history of Dunhuang [71]. Following several public discussions at international conferences of Dunhuang studies in the 1980s [72,73], Shi stated that Dunhuang art cannot be directly equated with historical facts and must be studied together with historical social conditions and cultural frameworks [74]. Like Ho, Shi believed that many visual images in Dunhuang murals are more than simple, straightforward reflections of the social lives of those who produced them; they were expressions of “the artistic and religious imaginations” of Chinese Buddhist artisans [75]. Marxist historical materialism, which had impacted the way in which first-generation scholars of socialist China criticized the limitation of Dunhuang Buddhist art, still lay in the root of the 1980s art criticism as exemplified by Shi.8 But in the Reform and Open-Door era (1978–), scholars adopted a more sympathetic view of the cave makers and were open-minded about learning about social realities and artistic expression from Dunhuang Buddhist art. Since the 1980s, Dunhuang Buddhist art of the later periods, especially the Guiyijun period, has piqued scholarly interest and has been studied from the perspective of the humanist spirit [76] and the local society of Dunhuang [77,78].

In this light, the cave temples obtained another reason to be studied as real existence: their spaces were the production—in French humanist geographer Henri Lefebvre’s use [79]—of human activities and social interactions. This approach was adopted by a series of scholarly works on the construction history of the Dunhuang caves, featuring Dunhuang scholar Ma De 馬德’s A Study of the History of the Dunhuang Mogao Caves (Dunhuang Mogao ku shi yanjiu 敦煌莫高窟史研究) [80]. By analyzing the changing cliff face, the studies introduced the notion of spatial interactions—despite being only at the scale of the cave complex—among cave construction of different periods. By reading the “merit records” (gongde ji 功德記) of the caves, commemorative texts praising the patrons’ merit in a construction project, the studies demonstrate how social interactions took place on a stage comprising the merit caves, caves constructed for the owner to accumulate religious merit [81]. Based on ownership, the merit caves could be further categorized as “family cave [of the laity]” (jia ku 家窟), “eminent monk’s cave” (gaoseng ku 高僧窟), “monastic communal cave” (sengtuan ku 僧團窟), “[Buddhist] society’s cave” (Shetuan ku 社團窟), and “government’s cave” (guan ku 官窟), among others [80] (pp. 331–360). Due to the interest in social history and the available historical sources, these studies dove deeper into the family caves of the prosperous clans than in any other kind of merit cave. A fruitful outcome has been the nuanced reading of the renovation of preexisting caves, a prevailing mode of construction activity at Mogao since the tenth century. Because the cave temple of an ancestor was treated as part of the family’s foundational work, the renovation was read as a necessary means of continuing the family tie [82]. Thus, prosperous clans in Dunhuang have been recognized as the major agents for the sustainability of cave construction at Mogao [83,84,85]. Nonetheless, a social art history can explain only part of the story of the Guiyijun-period caves; to better understand the changes in cave architecture, such as the significant moves in cave design that Shi Zhangru observed, one must also consider the intrinsic factors of the religious work of architecture [38].

3.5. Inquiries About the Relationships Between Image and Site, Cave and Stand-Alone Architecture

Understanding the cave temple from the perspective of architectural space is not easy because many norms of architectural studies are not applicable. The cave temple has been well recognized as a kind of Buddhist architecture, but its spatial logic is believed to be inherently different from its stand-alone counterparts. For instance, while Ho attempted an exhaustive study of the Buddhist monastic architecture of Sui and Tang China, he did not include the rock-cut cave temples. Ho’s dissertation aimed to reveal the spatial conception of architecture, but a spatial conception foreign to that of the architecture that had been conceptualized to inhere in the cave. Ho’s approach to spatial conception can be traced back to Swiss architectural historian Sigfried Giedion (1898–1968), who conceptualized the history of Western architecture as a succession of three concepts of space and saw in the cave a concept of space prior to the advent of architecture [86,87]. Subsequently, Japanese architectural historian Mitsuo Inoue 井上充夫 (1918–2002) applied the lens of spatial conception to inspect the developments of Japanese architecture, customizing a counterpart of Giedion’s narrative in East Asian history [88,89].9 While Giedion valorized the pre-architectural concept of space that the prehistoric cave displays to be a potential alternative to architecture as a human habitat, he regarded the cave as incomplete as an architectural space. The caves, in Giedion’s eyes, “possess an interior... but they have no exterior [87] (p. 526)”, and Inoue similarly saw a cave dwelling to be “nothing but interior space [89] (p. 3)”.

Although the Buddhist cave temple differs from the prehistoric cave dwelling by geometric form, visual decoration, and ritual function, the conception of them as essentially an “architecture of the interior” has remained intact. Architectural historian Wei-Cheng Lin has offered a critical review of this accepted idea and recent modifications [90], yet the interior-outdoor problem still awaits further investigation. The accepted idea is not based just on the fact that the main chamber of a Buddhist cave is physically an interior space but also on the fact that the form and decoration of the chamber, especially those built in the Tang or later periods, resembled the interior of a temple hall. Art historians have also alerted us that a cave temple would not fully adopt the functionality of Buddhist monastic architecture because the former is usually darker and smaller than a stand-alone temple [91,92]. Scholars of Dunhuang history have also pointed out that the Mogao caves did not function like a Buddhist monastery, which would be a courtyard complex comprising multiple halls, pavilions, and monastic dwellings ([80], p. 203, [93]). But even if the space of a cave was originally pre-architectural and incomplete, would anything have been changed when the cave was adorned by the images of architecture that present a complete set of architectural concepts of space?

Inquiries have been raised about the mutual relationship between the image of architecture and the actual architecture, independent of the cave specifically. The best-known case of image-inspired architecture is the Japanese Pure Land garden, which is accepted as having been directly influenced by the architectural imagery in Pure Land paintings [94,95,96]. The image of Pure Land architecture borrows compositional principles and building types from earthly prototypes such as monasteries and palaces, whereas the Pure Land garden’s compact composition, scale contrasts between the hall space and the decorative structures, and enriched skylines of the roofs are believed to be innovations that express the picturesque beauty of the visionary architecture. In contrast, the cave temple’s physical juxtaposition with the visionary architecture paradoxically makes its conceptual correlation less apparent, owing to its distinctive material construction systems and varied ways of defining space. The aspiration to extract information about Buddhist monastic architecture from cave architecture has existed in the minds of architectural historians since Xiao, but the triangular relationship among the image of architecture, the cave architecture, and the stand-alone architecture they represent remained unstudied until more comprehensive approaches to Dunhuang cave art emerged in the twenty-first century.

4. The Third Phase (Since 2001): New Materials, Perspectives, and Technologies

4.1. Blooming Publications, Correlated Visual Materials, and Surveys of Built Environment

Since the turn of the twenty-first century, the visibility of Dunhuang cave art has drastically increased in both traditional publications [97,98] and digital databases [35,99,100,101]. Among them, the most comprehensive study of the cave architecture of Dunhuang was conducted by two Dunhuang Academy researchers, Sun Ruxian and his eldest daughter, Sun Yihua 孫毅華. Having deeply engaged in the architectural conservation of the Dunhuang caves for many decades, the Suns gained unparalleled first-hand knowledge of the images of architecture, the cave architecture, and the cliff site. Their knowledge was condensed into two volumes of a twenty-six-volume photo album of the Dunhuang caves produced by the Dunhuang Academy, one on the image of architecture [9] and the other on cave architecture [10].

This bipartite division followed Xiao’s theoretical framework, yet an enriched reading of the cave architecture surfaced in the Suns’ work. First, their discussion of cave types took into consideration the broad spectrum of cave sizes and the constructive decoration that enhances the architectural imagery of a cave. This move added to the abstract concept of cave type certain spatial and visual features perceivable by a viewer.

Second, they began the exploration of the possible relationship between the image of architecture and the historical architecture in Dunhuang. For instance, they first recognized the resemblance between a Song-period flowery pagoda (huata 花塔) in the Daquan valley (Figure 6) and the pagoda image in a Xixia-period (1038–1227) mural painting of the Yulin caves (Figure 15) ([9], p. 260, [10], pp. 230–231).10 In another instance, Sun Yihua combined the appearance of the temple hall in Dunhuang murals and the interior of the central-pillar cave to reconstruct a type of pagoda-centered temple that might have existed in northwest China around the Northern Wei period (386–535) [102].

Third, as early as the 1950s, Sun Ruxian invented a new way to document the total visual form of the cave interior, considering both the shape of the volume of void space and the decorative surfaces wrapping around it. His cave renderings integrate the cave type, polychromic statues, and mural paintings into a coherent representation of space (Figure 16). Derived from preexisting templates of archaeological drawings of the cave form and trace-copy replication of the mural painting, this kind of rendering was revolutionary; it presented the visual contents of a cave in its entirety and in place, thus facilitating further investigation of the architectural and pictorial programs of the cave side by side. The principle continues to guide the digitization, digital conservation, and presentation of Dunhuang cave art in the present day [103].

The Suns’ general approach aligned with Xiao’s, as their discussion of the built environment was subordinate to the topological studies of cave architecture, and their purpose of incorporating the ancient architecture of Dunhuang was to prove the realistic aspect of the image of architecture. Nevertheless, the Suns’ studies, in fact, took the first step to blur the boundaries between the pictorial and the actual architecture of Dunhuang.

As surveys and studies of the built environment of the Mogao caves continue into the twenty-first century, scholars gained new knowledge about its historical look. For instance, archaeologist Sha Wutian 沙武田 and Sun Yihua, who collected massive archaeological information about non-extant ante-halls and porches, revealed the prevalent distribution and formal patterns of the cave-front architecture [104,105]. Recently, architectural historian Zhenru Zhou took a further step in relating the site’s historical appearance to the religious imagination of Buddhist heavens, palaces, and pagodas [106,107]. In another instance, Dunhuang scholars recognized and investigated the intentional grouping of caves, especially a type of vertical composite comprising two or three caves and a cliff-top pagoda ([10], pp. 239–40, [108,109]). Among newly published surveys and studies of ancient pagodas adjacent to the cave sites [110,111], the most comprehensive work is a Dunhuang Academy researcher Guo Junye 郭俊葉’s monograph on the earthen pagodas of the Mogao caves [112]. These studies shed light on the comprehensiveness of a cave site’s historical built environment, which comprised caves, cave groups, and affiliated or stand-alone structures in the natural landscape.

Inquiries into the historical built environments of Mogao have been enriched not only by recent fieldworks conducted by Chinese scholars but also by the belated publication of materials collected by the early foreign explorers, featuring the Russian expedition to Dunhuang led by archaeologist S. F. Oldenburg [113]. Prior to Oldenburg, Silesian-Austrian geographer Gustav Kreitner (1847–1893) [114], Hungarian-British archaeologist M. Aurel Stein (1862–1943) [115], and French Sinologist Paul Pelliot [12], among others [116], had visited Mogao and published their findings. Oldenburg largely adopted Pelliot’s cave numbering and cataloging system; however, his team produced the most comprehensive and precise drawings of the Mogao complex as it appeared in the early twentieth century. This work was conducted by architect Viktor Sergeyevich Birkenberg (1890–1938) and topographer Nikolai Arsenyevich Smirnov (1896–1983) [117] (p. 74). The drawings include a 12 meter long watercolor rendering of the entire cliff face (Figure 17), alongside several continuous plan drawings representing most caves at Mogao. In addition, the Russian team systematically documented their source materials, including photographs of cliff sections, diagrams, and textual records of individual caves. This collection offers invaluable insights into the historical appearances of the Mogao complex, most traces of which have been obscured by cliff reinforcement projects in the second half of the twentieth century. Located in the State Hermitage Museum in St. Petersburg, these significant materials were eventually enabled to reach a broader audience following the publication of a six-volume catalog in China in 1997–2005 [13]. 11

Another member of Oldenburg’s expedition team, photographer and painter Samuil Martynovich Dudin (1863–1929), was among the pioneers who studied the historical site of Mogao. His studies, which were originally published in Russian journals in the 1920s [118], were translated into Chinese and published in China in 2006 [119]. This recent publication provides international scholars with new evidence regarding the complex’s multi-phase developments. Dudin keenly observed how the numerous caves had once formed several architectural ensembles in different periods, some of which extended horizontally and others vertically [119] (pp. 84–85). This observation echoes findings by Chen Mingda [42] (pp. 53–57), Ma De [80] (pp. 50–90), and Sonya S. Lee [120] (pp. 201–205). The long-forgotten and subsequent studies converge to illustrate that the architectural ensemble of Mogao is an ever-changing, living entity.

4.2. Developments of Scholarship on Visual Representation of Architecture

As published materials of archaeological sites in Dunhuang and elsewhere significantly increased, architectural historians enlarged their visual sourcebooks of early Chinese architecture. The classical approach to the Dunhuang Pure Land paintings as pictorial representations of architectural space continued to flourish in the current century. Studies on Pure Land architecture [121,122,123] and those on the visual representation method of the Pure Land images [18,124,125] have significantly advanced. But more and more architectural historians—Fu Xinian 傅熹年 [126], Wang Guixiang 王貴祥 [127], and Nancy S. Steinhardt [128] prominent among them—adopted an inclusive approach to gain information about building styles and layouts from both the plastic and pictorial images of architecture from the Dunhuang caves and elsewhere, just as they dealed with funerary artifacts and tomb murals.12

This approach was exploited to an extreme in Zhao Nadong’s 趙娜冬 recent studies of Dunhuang cave architecture, visual programs, and images of architecture [131,132]. Zhao integrated and reinterpreted the published materials in these categories to respond to a classical problem of Dunhuang architecture: how the caves could represent the spatial layouts of Chinese Buddhist monasteries. The basic strategy for processing information from various kinds of visual sources has been an abstraction; like plan reconstruction for Pure Land images, the cave space with its main visual contents is translated into the layout of a cloister where the main ritual buildings are arranged in the cardinal directions (Figure 18). In this way, Zhao found information about spatial composition from a broader range of materials that had been unstudied from the architectural perspective, including the visual programs in a cave and the spatial layout of a cave group. She even claimed that a cave with a Buddha niche and two Pure Land scenes carries the spatial imagery of a monastery complex comprising three rows of courtyards, one central and two flanking [132] (p. 183). Although Zhao might not have overstated the cave design of a “total space” comprising “real and virtual spaces [132] (p. 184)”, her abstraction of both spaces ruled out the possibility of examining the embodied viewing experience of the total space. The dilemma is that as long as one looks for the architectural concept of space, one sees no more than symbolic and virtual space in the cave.

4.3. A New Perspective: Space as an Analytic Tool

A potential solution to the dilemma lies in a more comprehensive understanding of space, a critical term and analytic tool in the field of art history in the last three decades. In conjunction with the “spatial turn” in art history [133,134], many studies of Dunhuang Buddhist art became more conscious about visual programs, visuality, and religious function in the cave space [90,135,136,137,138,139]. The Buddhist sacred mountain, a natural site that inspired cultural imagination and a historical place where building activities occurred, has also been examined from art-historical perspectives. For instance, Wei-Cheng Lin [64] and Nachiket Chanchani [140] studied the dialectical relationship between Buddhist sacred mountains and temple architecture in East and South Asia, giving a new impetus to the classical inquiry of architecture, landscape, and religious culture. Cary Y. Liu also investigated the fundamental connection between Mount Sanwei and the Mogao caves that have underlain the latter’s architectural principles [141]. Sonya S. Lee examined the cave sites in southwest China from the perspective of sustainability, an emerging issue for ecological art history [142]. By placing Buddhist art and architecture in their spatial and historical contexts, such scholarship demonstrated that the decorated cave temples could have facilitated a “dialectical transcendence in real space” ([142], pp. 5–7, [143]).

Art historian Wu Hung, building on his studies of Dunhuang art over thirty years, articulated in a recent monograph, Spatial Dunhuang: Approaching the Mogao Caves, the most thorough discussion by far of space in the Mogao caves [144,145]. Wu Hung’s discussion provides three starting points to reconsider what space means for a cave site. First, the Mogao caves can be observed in spatial contexts of five levels, including (from large to small) the historical place of Dunhuang, the evolving entity of the Mogao cliff, the interior spaces of the caves, the sculptural and pictorial programs inside a cave, and the pictorial space within a mural painting ([144], p. 291, [146]). While the spatial contexts of each level are discussed as a stand-alone topic in Spatial Dunhuang, the implied subject that made the spectrum of space continuous is the human experience of making, using, and contemplating the spaces. Second, a subjective experience-based understanding of the visual space of a cave—the synthesis of architectural, sculptural, and pictorial spaces—can be analyzed in a scientific manner. Using the human body as a reference point, a researcher can investigate the scale of a cave relative to its visitors and the possible sequence of approaching the visual programs through their movements (Figure 19). The spatial property of the cave, which structures the movements of the humans and their relationship to divinity, is thus examined. Third, to study the relationship among multiple caves, Spatial Dunhuang defines a series of “originary caves”, which could be subdivided into “unique caves” and “model caves” according to their legacy at Mogao [144] (pp. 293–294).13 From this perspective, instead of being read as an “authentic visual sequence” of Chinese art and architectural history [147], the successively emerged paradigms at Mogao could be read as a symphony of dialogs among caves, even those of distant construction times. Although Wu’s investigation focuses on the individual cave, his strategy, an “analysis of the cave space”, provides a methodological foundation for understanding the comprehensive built environment of the Mogao cave complex in its own right. That is, the concept of space is not limited to the formal logic of architecture but also considers human experience and historical sensibilities.

4.4. Digital Technologies for Research and Display

Some other areas of study that have displayed the potential to redirect the discourse of real and virtual spaces are digital art history and art curation. In the last decade, computer-aided design technologies for processing architectural images in Dunhuang mural paintings have been significantly advanced 14, as have virtual reality (VR) [148], augmented reality (AR) [149,150], and digital fabrication technologies for replicating and re-creating Dunhuang cave art.15 As early as 2012, new media artist Jeffrey Shaw employed a stereoscopic projection theater or iPad screens to create immersive or interactive ways of exploring the exquisite early-Tang murals inside Mogao Cave 220 [151,152]. With the assistance of digital technologies, exhibitions have brought more vivid, stereoscopic images of the Dunhuang caves to museum-goers all over the world, including those in Hong Kong and Los Angeles.

Computation has allowed not just artistic and scientific simulations of the structure of the caves [153] but also creative manipulation of the digital assets of their visual contents [154]. With these digital tools, architects and computer scientists have designed and digitally constructed a few representative building complexes in Dunhuang Pure Land paintings [155,156,157]. For instance, the Pure Land architecture in Mogao Cave 217, whose layout had been reconstructed by previous scholars, was re-created by digital artist Ning Yuhang 寧煜航 into a virtual experience of constructing it block by block (Figure 20) [158].

While new visual technologies have helped visionary architecture in the pictorial space look real, they have also complicated the concept of reality associated with the material thing in the actual space. The exhibition A Thousand Years of Construction: The Beauty of Dunhuang Architecture [Qiannian yingzao: Dunhuang jianzhu zhi mei 千年營造——敦煌壁畫中的建築之美] at the Exhibition Center of the Dunhuang Academy since May 2022 presents a dialog between the real and virtual spaces. In an art installation, a walk-through video of a Pure Land building complex is projected on the rear wall of a cave-like showcase. The wall bearing the animated Pure Land scene stands as a backdrop of a physical model of another Pure Land building complex (Figure 21). The juxtaposition of the two architectural images, one alluding to an enterable space behind the wall surface and the other making that kind of space present in a miniature scale, questions whether a line can be drawn between the virtual and the real. Technically speaking, a clear distinction between real and virtual can be made about material things and their digitized replication [103]. Yet the mind’s eye may willingly ignore that distinction when spatial imagination is stimulated. The two models’ formal correspondence and complementary visuality enhance the viewer’s illusion that the architectural image has been “enlivened” and that it has “jumped out” from the pictorial plane into real life [159].

To take a step further, a visual object, even if it represents, imitates, or symbolizes other space, has its own place in this visual milieu. In this case, the physical and digital models are parallel existences modeled after the Pure Land images, but as long as they are placed in one space, they have visual resonance with each other. Echoing the conception of image (xiang 像) as an analogue in ancient China [160], these parallel existences are as effective as their counterparts in re-creating a spatial experience of the original, which might have existed only in mind. The nonextant ancient Chinese architecture, a phantom in studies of Dunhuang architecture, is one such idea of “the original”. Although it is always meaningful to trace that origin, its analogues—the image of architecture and the cave architecture—might not provide any straightforward path. Nonetheless, if their efficacy in space-making and placemaking is recognized, these Dunhuang materials can serve as surrogates for actual buildings, as the “best, most faithful, and most valuable materials” for architectural studies [3] (p. 2).

5. Reflections: Toward a Fuller Understanding of Space

5.1. Achievements and Limitations of Current Scholarship

Architecture-related materials from the Dunhuang caves have been proven important in the studies of Chinese art and architectural history, Dunhuang caves, and local history. Classical studies of Dunhuang architecture have sought to detect formal features of stand-alone architecture, especially Buddhist, from the images of architecture in mural paintings, the architectural typology of the cave chambers, and the timber-framed construction of the cave-front porches. They have revealed the diversity of architectural typologies and compositions in ancient China and along the Silk Road, confirmed the evolutional steps of the timber-framed construction system, and participated in deciphering the making of a religious landscape.

Meanwhile, a majority of scholarship has led to two accepted ideas about cave architecture and the way to study it. One idea is that the cave is an “architecture of the interior”; the other is that the main value of the cave for the studies of architectural history lies in its representation of something else. While both ideas have been reconsidered and modified in recent years, they still underlie architectural studies of the cave temples. A new set of ideas of cave architecture must be established before new knowledge of Dunhuang architecture is acquired.

This new set of ideas must go beyond the persistent focus on the image of architecture. As conventionally accepted, the images of architecture represented in mural paintings are more accessible and diverse than the actual architecture of caves at Dunhuang and, therefore, have raised more scholarly interests. Most of the building types represented in Dunhuang murals, except for some minor types of vernacular architecture, have been considered by Liang Sicheng and Anneliese Bulling in the 1930s and were systematically studied by Xiao Mo in the 1980s. More recent studies of architectural images are fine-grained and technically enhanced, producing knowledge within the classical framework established by Xiao.

In comparison, architectural studies of the actual caves have, to a much lesser extent, explored the potential of the rich materials. Surveys of the actual architecture of the Dunhuang caves and their affiliated structures require more fieldwork and permissions, more comprehensive surveying technologies, and better skills in processing various kinds of data. However, such prerequisites should not be excuses for ignoring this inadequately studied dataset. First of all, previous investigators—featuring the Russian expedition team and Shi Zhangru—and the Dunhuang Academy researchers—Sun Ruxian, Sun Yihua, Pan Yushan and Ma Shichang, and Guo Junye among others—have discovered and published more materials than what architectural specialists conventionally look at. Furthermore, developments and popularization of technologies for processing visual and spatial data have stimulated interdisciplinary collaborations, enhanced scholars’ capabilities, and aroused the public’s interest in exploring the cave site. Most importantly, only with knowledge of the actual architecture can one fully understand the conception, perception, and imagination of space that the image of architecture was made to shape.

5.2. Four Interrelated Aspects of Space

Space is not just about the physical dimensions and shapes of a void volume. As humanist geographer Yi-Fu Tuan 段義孚 (1930–2022) has theorized, space can be approached as a multiplicity of mental constructs that all rely on the interaction between the human body and its environment, whereas place is space enriched with the experiences and knowledge of human beings and given social and geographical locations [161]. From the perspective of human experience, the cave temple, as exemplified by those at Mogao, is a unique kind of architecture that encompasses the natural, the social, the lived, and the symbolic spaces.

The natural and social spaces allow us to inspect the relationship between nature and artifice in the production of this unique type of architecture. The natural space, referring to the Mogao cliff and the Daquan valley between Mounts Sanwei 三危 and Mingsha 鳴沙, is an important factor that the cave constructions must have considered. The social space, meaning that any cave is a social production, denotes the decisions made by the cave makers during the construction, renovation, and maintenance of a cave.16 From this perspective, apart from treating the Mogao caves as representations of architecture, an architectural study can also approach it as an actual locale subject to the need to deal with real architectural problems, including siting, arrangement of solid and void spaces, façade building, and maintenance and renewal of the built environment, among others.17

Inseparable from the problems of real architecture is the religious imagination of architecture. The lived and symbolic spaces help us understand the mutual relationship between actuality and virtuality in the spatial art of the cave. The lived space refers to the space that is perceivable, usually through ritual activities that require the worshiper to move through the space. The symbolic space refers to an idea of space in which the lived space is a sign, a representation, or an imitation. It is usually conveyed by the optical forms in a cave and relies on the viewer’s contemplation. The mutual relationship might be explored in terms of two aspects. On the one hand, since a Buddhist sacred space is usually conceived as a dwelling place of a deity, often represented by a main icon in the space, the lived space has an imaginary aspect. The viewer normally conceives that the space is not designed for him or her but for the deity, and a consecrated space must be enlivened from the point of view of the deity. On the other hand, it is a space’s actuality that enables its imaginary aspect to exist in collective memory. It should not be ignored that, more often than not, both modern and medieval beholders have been compelled to see the architectural imagery provided by the cave. In other words, the virtuality of a cave—“a certain force in having the effect of what they are not in fact”, as defined by art historian David Summers [163]—is a sharable experience of its lived space.

The four aspects of space, among others, provide new lenses through which one may investigate the Dunhuang caves. These lenses have the potential of rediscovering classical scholarships, integrating related-yet-traditionally-excluded materials, applying present-day technologies and approaches, and thereby reviving the almost hundred-year-old field of Dunhuang architecture studies.