**1. Introduction**

During the twentieth century, a number of miniature stone pagodas and hundreds of fragmented pieces were discovered at monastic sites and hoarding pits located in northern China (Figure 1). Among them, a particular group comprises pagodas formed by a series of cubical or trapezoidal stone blocks that are once stacked up in diminishing sizes (hereafter referred to as "stacked pagodas," or "stone blocks" if only a single block is under discussion). This group of pagodas is excavated primarily in the Nannieshui 南 涅水 County of Shanxi 山西 province (Figure 2), and several sites in Gansu 甘肅 province (Figure 3). According to dedicatory inscriptions, as well as styles of relief carvings on these stone blocks, they were commissioned over the course of the sixth century, when northern China was successively reigned by the Northern Wei 北魏 (386–534), Eastern Wei (534–50), and Western Wei (535–57), Northern Qi (550–77) and Northern Zhou (557–81). Historically, eastern Gansu was named "Longdong" 隴東, meaning "to the east of the Mount Long". Once a center of Buddhism, the region is home to several cave-temple sites and numerous Buddhist statues and steles that date to the Northern dynasties (Gansu Sheng Wenwu Gongzuodui, and Qingyang Bei Shiku Wenwu Baoguansuo 1987; Cheng 1998; Cheng and Yang 2003; Dong 2008; Song 2009; Wei and Wu 2009; Gansu Beishikusi Wenwu Baohu Yanjiusuo 2013; Zheng et al. 2014).<sup>1</sup> Nannieshui in Shanxi, although never a significant local center in history, is located on the path connecting major political centers in the sixth century (Guo 1959, 1979; Shanxi Sheng Kaogu Yanjiusuo 1994, pp. 313–18; Zhang 2005, pp. 51–68).

Besides excavation reports, and several catalog entries, there is almost no extensive scholarly discussion of the phenomenon of making pagodas by stacking stone blocks. These pagodas' display of rich Buddhist imagery on the surface, in contrast to the simplification of structural elements, raises pointed questions about religion, imagery, and architecture. Why were stacked pagodas made? How to understand their regional flourish in Shanxi and Gansu? How were stacked pagodas understood in their production, consumption, and veneration? How did they relate to the construction of pagoda buildings and other pagoda-centered Buddhist activities of the time period?

**Citation:** Zhao, Jinchao. 2021. Generating Sacred Space beyond Architecture: Stacked Stone Pagodas in Sixth-Century Northern China. *Religions* 12: 730. https://doi.org/ 10.3390/rel12090730

Academic Editors: Shuishan Yu and Aibin Yan

Received: 14 July 2021 Accepted: 24 August 2021 Published: 6 September 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Figure 1.** Map of Eastern Gansu, Shaanxi, and Shanxi province.

**Figure 2.** A Nannieshui pagoda in display. Early sixth century, Northern Wei. Stone. Nannieshui Museum. Photograph taken by the author.

The pagoda is usually considered the reinterpretation of the hemispherical stupa, but ¯ features a tall, multistory tower-like body instead. In the third century BCE, Buddhism adopted the mound structure to house the relics of the Buddha or to mark the places consecrated as sites of his acts, constituting the earliest presence of stupa. ¯ 2 Despite the scarcity of material remains dated prior to the fifth century, literary sources record that the earliest stupas were produced in China no later than the second century CE, along with ¯ the eastwards spread of Buddhism.<sup>3</sup> From the third century, Buddhism rapidly filtered into Chinese society through increasing contacts with Central Asian Buddhist kingdoms, translations of Buddhist texts into Chinese, and constructions of Buddhist monasteries (Zürcher 1972; Seishi 2017). The tower-shaped and multistory type of stupa appeared and ¯ later became the predominant form in East Asia. Modern scholars usually referred to this type as "pagoda" in distinction from the hemispherical stupa. Both types are generally ¯ referred to as *ta* 塔 in Chinese scholarship.

**Figure 3.** Zhuanglang Pagoda. Gansu Province. Early sixth century, Northern Wei. Gansu Provincial Museum. Photograph courtesy of Wang Xiaoshu.

The period in the following fifth and sixth centuries was marked by the conquest of northern China by non-Chinese regimes. Buddhism continued to flourish regardless of the political chaos and social upheaval. Both imperial and ordinary patrons sponsored the construction of pagoda buildings. Pagoda and stupa imagery in reliefs and murals burgeoned ¯ in Buddhist cave-temples and on Buddhist statues and steles. Surviving historical texts from the time, especially *Luoyang qielan ji* 洛陽伽藍記 (A Record of Buddhist Monasteries in Luoyang) by Yang Xuanzhi 楊衒之, and *Shuijing zhu* 水經注 (Commentary on the Water Classic), compiled by Li Daoyuan 酈道元, record a myriad of Buddhist pagodas being erected within the precinct of Buddhist monasteries in Luoyang (*Luoyang qielan ji jiaoshi* 1963; *Shui jing zhu* 2007). An upsurge of archaeological excavations of pagoda foundations of the Northern dynasties since the second half of the twentieth century CE has brought to light information on pagoda constructions dated to this period, such as the *Siyuan fotu* 思 遠佛圖 at Pingcheng 平城 (present-day Datong 大同, Shanxi Province), *Siyan fotu* 思燕佛圖 at Chaoyang 朝陽, Pagoda of the Yongning Monastery 永寧寺at Luoyang 洛陽, the Taihe Pagoda at Dingxian 定縣 County, Hebei, and pagodas in temples located at the City of Ye 鄴, Hebei (Hebei Sheng Wenhuaju Wenwu Gongzuodui 1966; Xia 1966; Dingxian County Museum 1972; Du 1981; Xu 1994; Zhongguo Shehui Kexueyuan Kaogu Yanjiusuo 1996, 2010, 2013; Su 2011).

In comparison to their Indian predecessors, surviving Chinese counterparts exhibit much greater diversity in forms, functions, and contexts. The earliest surviving group of stone pagodas/stupas includes the thirteen miniature pieces that were discovered in the ¯ last century in major Buddhist sites of northwestern China.<sup>4</sup> They share an almost identical

design of domed top and three registers on body. There are six bear inscriptions dated to between 426 and 436, when the region was ruled by the Northern Liang 北涼 (397–439).<sup>5</sup> However, the form of the Northern Liang group did not achieve currency in the following period. Instead, around twenty multilevel stone pagodas became predominant, which have a square, tower-like body of three, five, seven, or nine stories, adorned with roofs, brackets, and columns imitating wooden structures, and feature a central niche on each side of each story (Figure 4). For instance, the Cao Tiandu 曹天度 pagoda consists of a square pedestal, a nine-level tower-like body, and a top, with parallel roof rafters rendered on each roof tier beneath an imitation ceramic-tile roof.<sup>6</sup> Other works similar to the Cao Tiandu Pagoda from the fifth and the sixth century were usually discovered near Pingcheng, Shanxi, and in western Gansu.

**Figure 4.** Cao Tiandu 曹天度 Pagoda. Shuozhou, Shanxi Province. 466, Northern Wei. Stone. H. 211.7 cm (with chattra top). National Palace Museum, Taipei (*Chattra* preserved in the Chongfu Monastery, Shanxi). Photograph courtesy of Tu Shih-yi.

In contrast to these typical multistory stone pagodas discussed above, stacked pagodas of the sixth century, which are formed by a series of cubical or trapezoidal stone blocks that are stacked up in diminishing sizes, become particularly intriguing in terms of the absence of any architectural elements (see Figures 2 and 3). On surviving stacked pagodas, there are barely any pieces that display components imitating wooden structures.<sup>7</sup> However, we know that these sixth-century stone pieces were made as pagoda, according to inscriptions and the vertical rise of their multilevel structure. Several bear inscriptions that refer to themselves as *shi fotu sanjie* 石佛圖三劫 (three-story stone pagoda), or *shi futu sanji* 石 浮圖三级 (three-story stone pagoda).<sup>8</sup> *Futu* 浮屠/浮圖, and in some cases *fotu* 佛圖, are synonyms of *ta* in history.<sup>9</sup> The emphasis of the three levels in these inscriptions, indicating the multilevel structure of the stacked pagodas, is universally used in epigraphs describing the dedication of pagoda buildings.

Furthermore, each surviving stone block of stacked pagodas displays rich Buddhist imagery on each of the four sides, revealing that imagery plays a quintessential role on these pagodas. However, the very act of adorning stacked pagodas with imagery and the arrangement of these images have not ye<sup>t</sup> been discussed. Previous scholarship focused primarily on examining individual depictive scenes (Guo 1959; Guo 1979; Cao 2011). A comprehensive evaluation of the iconography and style of images depicted on stacked pagodas is much needed to better understand the mechanism of producing stacked pagodas.

The pondering of the mechanism of dedicating stacked pagodas is particularly important in view of the broader historical context of constructing pagoda buildings and commissioning stone pagodas in fifth- and sixth-century China. Miniature stone pagodas from the time, often studied only as imitations of actual wooden buildings, and as evidence to reconstruct the look of real buildings that barely survived, were rarely examined for their own instances. Major scholarship in art and architectural history has focused mainly on reconstructing a linear development based on formal analysis of the stupa/pagoda's ¯ structure.<sup>10</sup> This interest in an evolutionary narrative of pagoda buildings understands surviving visual materials of stupas and pagodas, including miniature pieces and the picto- ¯ rial representation in reliefs and murals, primarily as evidence for reconstructing monastic architecture, resulting in limitations in further inquiry of other valuable aspects. Despite the contribution to the advancement of the field, this evolutionary approach becomes problematic in two aspects in the study of non-built forms: the "origin" narrative, and a fabricated evolutionary scheme. Meanwhile, research in the field of Buddhist studies lends itself to abstract discourses in history and doctrine, rarely contributing to analyses of specific images or objects.

While I am indebted to the previous scholarship conducted on the architectural aspects of pagodas, this essay aims to further explore how stone pagodas, particularly stacked pagodas from sixth-century Gansu and Shanxi, display their function and significance beyond their architectural form. Why were they made in the first place? What do we make of the images lavishly adorning their surfaces? How were they situated in the artistic, historical, and religious milieu of the time? How were they related ye<sup>t</sup> differentiated from the contemporaneous practice of building pagodas and dedicating Buddhist sculptures?

The first section of this study offers a survey on the style, iconography, and patronage of surviving stacked stone pagodas. The second section examines the regional feature of making stacked pagodas in eastern Gansu and Shanxi, respectively, and proposes an eastern Gansu origin given the historical context. It also highlights the local reception and promotion of hybridity, which is exhibited by the integration of styles, motifs, and pictorial programs of diverse origins in both regions.<sup>11</sup>

The third section pinpoints the absence of architectural elements imitating wooden structures on stacked pagodas, which is in contrast with the rich Buddhist images covering every corner of their surface. In the fourth section, I further examine the broader historical milieus of producing stone pagodas in preceding periods, showing a successive tradition of communicating Buddhist teachings through imagery depicted on the surface of stone pagodas. On stacked pagodas, however, the well-configured pictorial programs featured on fifth-century stone pagodas are dissembled, with each individual image exhibiting independent significance. Such an abstraction and distortion of their shapes exhibits a tension between form and meaning.

The fifth section explores factors that shaped this new emphasis on individual images on stacked pagodas, arguing for the ignored correlation between the configuration of images and the organization of patronage, which deeply shaped the way how stacked pagodas were commissioned and venerated. According to surviving inscriptions, an individual image located on one side of a stone block from Shanxi, instead of the stone block or the stacked pagoda, becomes the targeted unit to be donated by a single patron. Following this emphasis on the perception of stacked pagodas among the local community, the sixth section studies the reference to pagodas in surviving inscriptions, and discovers a transition from *futu* 浮 圖 (pagoda) to *xiang* 像 (image). I contend that it is the growing

emphasis on individual image, which was formed by the very making-process of these stacked pagodas in module, that gradually altered the perception of them from pagodas to images.

#### **2. Stacked Pagodas from Gansu and Shanxi**

## *2.1. Gansu*

There are two primary geographical loci for the discovery of stacked pagodas: Pingliang 平 涼 Prefecture in eastern Gansu province and Nannieshui at Qin 沁 County in southeastern Shanxi province.<sup>12</sup> Eastern Gansu in its modern administrative division includes two sub-areas: Qingyang 慶 陽 and Pingliang. Once a center of Buddhism, the region is home to several cave-temples, such as the Northern and the Southern Cave-temples 南北 石 窟 寺 at Qingyang, Wangmu Palace 王 母 宮, and Luohan Cave 羅 漢 堂, and numerous Buddhist statues and steles that date to the late fifth century.<sup>13</sup> The development of Buddhist art in Pingliang relates closely to two centers at the time. The first is Chang'an 長 安, Shaanxi 陝 西 province, which is adjacent to eastern Gansu, and has been a political and cultural center since the fourth century. Another center is the Maijishan 麥 積 山 Cave-temples at Tianshui 天 水, located to the southwest of Pingliang.<sup>14</sup> Overall, Pingliang Buddhist art exhibits distinctive regional features but also possesses a blend of stylistic traits that had already circulated in the adjacent area of Chang'an and Tianshui, as well as Pingcheng, the Northern Wei capital city and art center of the fifth century.

Among the dated stacked pagodas in eastern Gansu, the earliest one was commissioned in 503 (Figure 5).<sup>15</sup> A square stone block from this pagoda was discovered in 1982 among over forty similar ones, as well as Buddhist steles and statues, at the site of Chanfosi 禪 佛 寺 in the Caowan 曹 灣 Village, Kongtong 崆 峒 District, Pingliang. Three other stone blocks respectively bear dates of 514,<sup>16</sup> 518,<sup>17</sup> and 519. They are identifiable as parts of stacked pagodas, given their trapezoidal shapes, the varied sizes, and the traces of mortise and tenon on the top and the bottom. Each face of the stone blocks was carved with images of Buddhas, bodhisattvas, narrative scenes, or decorative motifs in recessed niches. On each side of the 503 block, there is a seated Buddha and two flanking bodhisattvas depicted within niche (See Figure 5). The Buddhas are dressed in monastic robes with symmetrical folds and sleeves with sharply defined edges splayed below the hands. The 518 Pagoda and other undated pieces from Chanfosi show a more diverse iconography. The Twin Buddhas motif (a pair of S´akyamuni and the Past Buddha Prabhutaratna), which was created in ¯ China based on the *Lotus Sutra ¯* , was becoming popular. Representing the Buddhas of the Present and the Past together, the motif underscores the notion that more than one Buddha can exist at the same time in the cosmos.<sup>18</sup>

Another group of Buddhist stacked pagodas was discovered in 1990 along with over 20 other pieces of statues in a hoarding pit located at Xiejiemiao 谢 家 庙, Huating 华 亭 County.<sup>19</sup> Three sets of stacked pagodas, each formed vertically by three stone blocks, are identified. All the three sets bear inscriptions: two sets were commissioned in the years 516 and 534 of the Northern Wei; and one set in 558, Northern Zhou (Figure 6).<sup>20</sup> The Northern Wei pieces exhibit hybridity of styles from different origins. On the 534 piece, the main Buddha image on one side dresses in a typical Han-style robe with sashes. The Buddha's right foot protrudes with the sole facing outside, representing a particular feature that is usually found in cave-temples of Longmen bsmi 龙 门 and Gongxian 巩 县 of the 520s near Luoyang.<sup>21</sup> Another Buddha triad depicted on the same piece renders petal-like edges of draperies in symmetry, exhibiting the local style of Gansu. Pieces of the second half of the sixth century instead show a more simplified iconography echoing other contemporaneous sites across the north, while their style is distinguished from the squarish style of other Northern Zhou sites.

**Figure 5.** Stone Block from Chanfo Monastery, Kongtong District, Pingliang City, Gansu Province. 503, Northern Wei. H. 44 cm, W. 33 cm. Pingliang Museum. Reprinted with permission from ref. [Gansu fojiao shike zaoxiang]. 2000, Zhang, p. 99.

A number of undated stone blocks discovered at Zhuanglang 莊浪 embrace dimensions and styles similar to the above-discussed cases.<sup>22</sup> A specific group was excavated in the courtyard of a Northern Wei Monastery, Baoquansi 寶泉寺, including one individual stone block and five that form a stacked pagoda (see Figure 3, hereafter the Zhuanglang Pagoda).<sup>23</sup> There are also several pieces found scattered in cave-temple sites near Zhuanglang. At the Wangmugong 王母宫 Cave-temple, a stone block in trapezoidal shape was once found on display. This piece, together with six others in situ, was brought to the cave-temple in the 20th century without much information of its original recorded provenance.<sup>24</sup>

## *2.2. Shanxi*

At Nannieshui, Shanxi, cubical stone blocks used to form stacked pagodas also survived in large number.<sup>25</sup> They were discovered in 1957 from several hoarding pits (see Figure 2). In 1990, another excavation led by the Archaeological Institute of Shanxi and the County Museum at Nannieshui discovered the foundation of a temple dated to the Tang Dynasty, as well as over 100 pieces of stone Buddhist statues of the Song Dynasty (Shanxi Sheng Kaogu Yanjiusuo 1994, p. 313). The relatively well-preserved condition of all surviving pieces without any apparent pattern suggesting deliberate breakage, in contrast to the broken nature of statues discovered in some other hoarding pits discovered in Hebei and Shandong, suggests that the Nannieshui sculptures were not damaged before their interment in these hoarding pits. Nevertheless, the lack of records of the first excavation

in 1957 makes it difficult to explore the nature of the burial and the original deposit of these sculptures.<sup>26</sup>

**Figure 6.** Pagoda from Xiejiamiao, Huating, Gansu Province. 558, Northern Zhou. Stone. H. 78.6 cm, W. (bottom) 26 cm. Huating Museum. Reprinted with permission from ref. [Gansu guta yanjiu]. 2014, Gansu Sheng Wenwu Ju, pp. 55–56.

Totaling 1373 blocks, each side of these cubic stones depicts Buddhist images, including Buddhas, bodhisattvas, guardian kings, narrative scenes from the Buddha's life story, as well as images of lay donors. Almost all the sculptures were carved from the grey, fine-grained sandstone that could be quarried locally. Based on surviving inscriptions, their execution dates range from 510, Northern Wei, to 1031, Northern Song 北宋. Most of the sculptures date to the sixth century, from the late Northern Wei through the Eastern Wei and the Northern Qi dynasties.

Previous scholarship on Nannieshui tends to introduce the materials among other Buddhist sculptures from southern Shanxi, highlighting the provincial style of the region but leaving space for more in-depth research. Yagi Haruo examines the depiction of intertwined dragons, the sun and the moon, and acrobatics among Nannieshui discoveries. Hinako Ishimatsu argues for a provincial style that has been shared by southern Shanxi and eastern Gansu and developed independently from the capital styles at the Yungang 云 冈 Cave-temples near Pingcheng and Longmen 龍門 near Luoyang. Gao Meng's dissertation on Nannieshui pagodas provides a survey without much analytical discussion in connection to contemporary Buddhist art (Yagi 2004, p. 84; Ishimatsu 2005, pp. 185–86; Gao 2012). In *Shanxi kaogu sishinian*, a four-period chronology is provided as an overview of the Nannieshui materials (Shanxi Sheng Kaogu Yanjiusuo 1994). However, this chronology has failed to thoroughly recognize the complexity of Nannieshui materials that is shown by the hybridity of styles and images with various origins and arranged in an array of ways.

The current study suggests a relative chronology for the Northern Wei materials based primarily on style, in combination with an extensive discussion of the change in iconography over the sixth century at Nannieshui. The proposed chronology divides Nannieshui materials into three groups: the Northern Wei group, the Northern Qi group, and the Tang and Song group.<sup>27</sup> The complexity of the Nannieshui materials lies primarily in the Northern Wei group, with multiple styles and rich iconography carved. The later

Northern Qi group shows stylistic traits that echo the idiom of contemporaneous Hebei and Shandong, which features a fuller body of the Buddha and a dressing mode that features thin draperies clinging to the body tightly. Those of the Tang and Song dynasties only constitute a small portion; the very fact that they were discovered along with those dated to the Northern dynasties in the same pits suggests that these works could not have been buried earlier than the tenth century, and those of the Northern dynasties might have been venerated throughout the five centuries in between.

Among the Northern Wei group, Buddha figures are modeled with slender bodies, squarish heads, and thick robes with draperies cascading on the throne, appearing to be ramifications of the late Northern Wei style popular in other areas. Nevertheless, one finds apparent regional variations between Nannieshui and eastern Gansu in their craftsmanship, the richness of decorative details in the background, and specific details with various origins. For instance, image no. 742 employs a typical Han style in rendering the sash of the Buddha's inner robe, with a rarely seen flower-like edge, as well as the thickly folded draperies that were popular at Luoyang and Majishan (Figure 7). Meanwhile, in the same scene, the Buddha sitting on the right protrudes the left hand to the other Buddha, exhibiting an idiosyncratic way of depicting the twin Buddhas motif of which one finds no examples beyond the area. In addition, the niche is surrounded by floral petals supporting Mani jewels half of which is cut off, two beings symmetrically arranged in between the petals, and one climbing onto the petal on the right, forming a design that is not found in other areas. The broad variety of details exhibited by Nannieshui pieces dated to the first half of the sixth century shows the strong individuality of each piece of work. Their variations sugges<sup>t</sup> the coexistence of various factors that are involved in modulating each individual work, and challenges the traditional approach that examines them as one group.

**Figure 7.** Twin Buddhas. On a Nannieshui Pagoda, scene no. 742. Early sixth century, Northern Wei. Shanxi province. Nannieshui Museum. Photograph taken by the author.

#### *2.3. Between Gansu and Shanxi*

It remains under-exploration about the relation between eastern Gansu and Shanxi in making stacked pagodas. While it is difficult to pinpoint the exact origin of the practice of stacking stone blocks to form pagodas, current evidence alludes to a greater likelihood of eastern Gansu origin. The earliest dated stone block is found in eastern Gansu, in the year 503, predating the Nannieshui pieces. Meanwhile, Lushui hu 卢 水 胡 (barbarians from Lushui), the ethnic group based at Jingchuan 涇 川 (Anding 安定 County in Northern Wei), eastern Gansu, has gradually migrated eastward to Shaanxi since the third century, constituting a major power in the area.<sup>28</sup> By the early fifth century, a series of rebellions were led by the Lushui hu residing in Shaanxi and eastern Gansu. Their fervent support of Buddhism is considered by scholars as a factor accelerating Emperor Taiwu's (r. 423–452 CE) decision of prosecuting Buddhism. Despite the failure of these rebellious actions and the elimination of the Juqu 沮 渠 clan who was behind most of the revolts, ordinary Lushui hu retained their stay in eastern Gansu and Shaanxi. Meanwhile, the Xue 薛 clan of the Hedong 河 東 Prefecture (present-day southeastern Shanxi), which lies in between the region where Nannieshui is located, Shaanxi, and eastern Gansu, maintained to be an ally of the Lushui hu through the fifth century, allowing for frequent traffic across Shanxi, Shaanxi, and Gansu (Liu 2005, 2008). The Xue Clan, originated from Shu 蜀 in present-day Sichuan, migrated to Hedong in the third century, and gradually became a predominating clan active in the area. The Xue clan retained its prosperity in the Hedong prefecture until the fall of the dynasty in 530s, and maintained contact with Lushui hu from the west along the way (Shanxi Sheng Kaogu Yanjiusuo 1994). Despite the absence of any stone blocks from the Hedong area, its geographical location at the crossroads connecting other parts of Shanxi with Shaanxi and Gansu, and the stability provided by the Xue clan during the Northern Wei, provided a path through which the practice of making stacked pagodas was transmitted in the north. Furthermore, numerous cubical stone blocks discovered in Shaanxi resembled the Gansu and Shanxi cases in size, style, and iconography (Abe 2001). The Shaanxi group's provenance in the Chang'an area—which is located right between eastern Gansu and southern Shanxi—connects Nannieshui and eastern Gansu, the two areas seemingly isolated from each other (Figure 8).

**Figure 8.** Stone block dated to 501, Northern Wei. Beilin Museum, Xi'an, Shaanxi province. Photographs taken by the author.

#### **3. Architectural Elements in Absence, Buddhist Imagery in Presence**

Among surviving stacked pagoda pieces from eastern Gansu and Shanxi, an intriguing phenomenon appears to be the lack of any structural components that imitate wooden building. In addition, the emphasis on adorning every side of pagodas with images is consistent of the fifth-century tradition. In the following discussion, I argue that the particular emphasis on pictorial programs adorning stone pagodas has already achieved prominence since the fifth century, showing the underlying mechanism of stone pagodas to generate significance and communicating with viewers beyond its architectural form. On stacked pagodas of the sixth century, structural components were further eliminated, leaving the veneration foci more focused on the images adorning pagoda surfaces, and showing a growing emphasis on images. Moreover, the relatively unified pictorial program of the fifth-century pagodas was also abandoned. Instead, images on stacked pagodas feature a more individualized and independent arrangement, constituting another piece of evidence to assert the importance of images.

To delve into the details depicted on stacked pagodas, a fallacy should be cleared at the outset. In several studies, a common approach is to study the vertical alignment of the images located on each of a pagoda's four sides, according to their current installation in museums (see Figures 2 and 3) (E 2011). It is natural to study a pagoda by looking into each of its facets, however, this approach is obviously not suitable for stacked pagodas, since the original sequence for arranging the stone blocks, if there is any, is lost. It is beyond any possibility to reach an affirmable reconstruction of the vertical connection among those cubic stone blocks as they were scattered even before being buried underground. The current logic of restacking these unearthed stone blocks relies primarily on their stylistic similarities. In other words, the way how stone blocks are displayed in museums should not be considered as their original arrangement, but an attempt to reconstruct. What we are certain about is the configuration of the four images located on the same cubic block. In the following discussion of images, any pictorial program is presented based on their content and their connection with the three other sides of the same block.

Taking the Zhuanglang pagoda as an example of the eastern Gansu group, images featured on each side of each story show a grea<sup>t</sup> variety of iconography, some of which enjoyed particular popularity (see Figure 4). In addition to Buddha triads, the most frequently depicted subject, scenes from the Buddha's Life Story (Ch. *fozhuan gushi* 佛 傳 故 事), appears to be a preferred topic. The Buddha's Birth and Bath are combined on the same pictorial space on one side of the fourth stone (Figure 9a).<sup>29</sup> On the lower half of the side, the space on the right is occupied by the scene that represents the Buddha's Birth as it follows the contemporary convention that depicts the baby Buddha emerging from the right flank of Queen May¯ a. Depicted in the middle of the space is the infant ¯ Buddha taking a bath; he stands naked on a three-legged stool, under a canopy formed by the nine-headed naga (serpent divinity translated into Chinese as a dragon), with two ¯ attendants pouring water from a vase. On the left we see the infant Buddha held by an ascetic astrologer invited to the palace to tell the future of the prince. The similar grouping of episodes from the Buddha's Life Story has been popular since the mid-fifth century in Chang'an and Pingcheng area.

Another episode from the Life Story of the Buddha is the Great Departure, which represents Prince Siddhartha departing the palace, acquiring importance for its highlighting ¯ of the historical Buddha leaving his princely life for a journey to seek enlightenment. In the scene on the first story of the Zhuanglang Pagoda, two standing figures and a horse in the foreground represent the moment when the prince is leaving with his steed (Figure 9c). Another figure raising hands in a building in the background shows the upset emotion of the prince's family in the palace.<sup>30</sup>

**Figure 9.** (**<sup>a</sup>**–**d**) Five-story Pagoda from Baoquansi, Zhuanglang, Gansu province. Early sixth century, Northern Wei. Stone. H. ~45 cm, W. 28–34 cm. Gansu Provincial Museum. Reprinted with permission from ref. [Gansu fojiao shike zaoxiang]. 2000, Zhang Baoxi, p. 25.

The Parinirva¯n . a scene is carved on the fourth story of the pagoda (Figure 9c). Referring to the Buddha's "death", the episode is represented with reclining Buddha with the head pointing to the left and feet to the right, and an accompanying group of mourners crying in the background.

An interesting question arises: why do the episodes of Parinirva¯n. a and the Great Departure appear independently while the other scenes are clustered in the same scene? A contextualized survey of precursors of these scenes shows that the two have gradually achieved independence since the late-fifth century for their respective teachings that constitute self-sufficient importance. The earliest Parinirva¯n. a image with a reclining figure in the center is found in Yungang Cave 11, reliefs within which dated to the 480s and 490s, as well as Phase III Yungang cave-temples.<sup>31</sup> Although the scene at Yungang is not prominent given its modest size and marginal location below a small niche, it stands out because of its separation from the sequence of the life story cycle. Similarly, the Parinirva¯n. a image on the Zhuanglang Pagoda is not aligned with other narrative episodes.<sup>32</sup>

The particular independence achieved by the image of Parinirva¯n. a and the Great Departure suggests a phenomenon of iconization, or the making of icons, which is also attested by the depiction of the A´soka story. Located on the third story of the Zhanglang Pagoda, the A´soka story represents the teaching that good karmic practices will lead to favorable reincarnation (Figure 9a). In the narrative, the Buddha and his disciple Ananda encounter several children playing outside during their trip. One of the children, in the hope of making offerings to the Buddha, takes a handful of soil and climbs upon another person's shoulder in order to reach the Buddha's alms-bowl. The Buddha accepts the earth and predicts that the boy would be reborn as King A´soka. In China, this story is commonly represented by three children making offerings to a standing Buddha. The boy who stands closest to the Buddha will become King A´soka. The story originally belonged to the category of avadana tales that correlates the virtuous deeds of the Buddha's past lives ¯ to subsequent lives' events. The story was accounted in *A´sokavad ¯ ana ¯* , which was translated into Chinese by An Faqin 安法欽 (active at Luoyang from 281 to 306), at the turn of the fourth century as *Ayu wang zhuan* 阿育王傳, and later by Sanghapala ¯ 僧伽婆羅 (460–524 CE) in 512 CE as *Ayu wang jing* 阿育王經.<sup>33</sup> The story was also found in *Xianyu jing* 賢愚經 (the Sutra of the Wise and the Fool), a collection of tales translated into Chinese in the fourth century.<sup>34</sup> Already shown in relief around the second century CE in Gandhara, the A´ ¯ soka story began to be depicted in the Phase II Yungang Cave 12 of the 480s, and further gained wide popularity in Phase III Yungang cave-temples and steles of the early sixth century (Strong 1983; Behrendt 2003, 2007; Brancaccio and Behrendt 2006).<sup>35</sup> The story is also found depicted on some of the Chanfosi stone blocks and cave-temple sites in eastern Gansu (Zhang 2000, p. 104, Figure 109; Gansu Sheng Wenwu Gongzuodui 1987, pp. 11–15). The A´soka story depicted on stone blocks follows the convention of depicting three children approaching a standing Buddha, while featuring additional elements including flying apsaras and two attendant bodhisattvas.<sup>36</sup>

All these narrative scenes, the A´soka story and those from the Buddha's Life Story, are also carved frequently among the Nannieshui group. The A´soka story at Nannieshui shows the Buddha standing, with his right hand touched by a child who is supported by another one (Figure 10). The Great Departure includes all the major elements we see in contemporaneous examples: a horse at his feet, and a tilting tree behind the prince (Figure 10). The composition within the square space reveals a possible influence from Yungang, as indicated by the modeling of the horse in high relief. The pensive Buddha image is shown sitting below a tree in the background and additional decorative elements carved in low relief surrounding the niche, including thatched huts that are usually associated with the venue of meditation, and Brahminic figures shown with knotted hair, naked upper bodies, and sometimes holding staffs. The highlight of the Buddha sitting in pensive posture among the Nannieshui cases distinguishes it from the depiction of the Buddha riding a horse in eastern Gansu. This regional difference might result from a growing importance of the pensive figure in the sixth century.<sup>37</sup>

The Parinirva¯n . a image, meanwhile, is represented in a completely new idiom at Nannieshui, which combines the typical scene showing a reclining Buddha side by side, with another scene that depicts a coffin (Figure 11). Both scenes feature a pictorial space formed by two intertwined tree crowns in the background. Such a combination reminds us of the renowned stele preserved at the Art Institute of Chicago (Lee 2010, Figure 1.1). The top register of the stele's reverse side renders two Parinirva¯n. a scenes almost identical to the Nannieshui case. This depiction of Parinirva¯n. a in two images is only found in sculptural remains dated to periods from the late sixth century. In her case study of the stele, Sonya Lee argues that the introduction of the coffin helps expand the temporal scope of the actions of mourning in depiction and establish a relationship of symbolic equivalence with the reclining Buddha (Lee 2010, chp. 1). Due to the survival of sculptures by chance, we do not know what the source is for this provoking invention of a two-scene representation of the Parinirva¯n . a. However, the provenance of the Chicago stele in southern Shanxi along the Fen River clearly indicates the circulation of this double depiction of a reclining Buddha and a coffin in the area.

**Figure 10.** Scenes no. 993–96, stone block from Nannieshui. Early sixth century, Northern Wei. Shanxi province. Photographs taken by the author.

**Figure 11.** Nirva¯n. a. Scenes no. 430 and no. 431. Stone block, Nannieshui, Shanxi province. Mid-sixth century. Nannieshui Museum. Source: (Cao 2011, Figures 6 and 7).

Meanwhile, the Northern Wei group from Nannieshui shows some idiosyncratic traits. Although the most commonly depicted images on a stone block include the seated Buddha, Maitreya, and the Twin Buddhas, we find numerous new elements that are distinctive from

the traditions flourished in other places in late Northern Wei, as well as unidentifiable subjects. In addition, the arrangemen<sup>t</sup> of images on these Nannieshui pagodas is arbitrary to a degree. To better examine the vast number of stone pagoda blocks surviving from the area, Gephi, the network analysis tool that uses the direction and frequency of links between nodes, is employed.<sup>38</sup> The nodes of the tool in this survey refer to imagery depicted on each side of Nannieshui stone blocks. It reveals that the A´soka story and the Great Departure usually are aligned horizontally with a seated Buddha, the Twin Buddhas, and Maitreya.

Overall, a survey of the narrative scenes in individual pictorial space on stacked pagodas from eastern Gansu and Shanxi reveals their prototypes developed in the late fifth century Yungang. The quick absorption of these individual narrative scenes on stacked pagodas shows recognition of their significance and popularity among the worshippers, and further alludes to a parallel correlation with steles and statues made around the same time. Each of these prototypes found at Yungang is also depicted independently from any other narrative scenes. This preference of Yungang tradition in eastern Gansu might relate to the historical context of the proliferation of Buddhist cave-temple sites at Longdong. An important site on the crossroads of trades and military campaigns connecting the Central Plains to the Hexi Corridor, the Longdong area has been kept close with the political center of Pingcheng. Dowager Empress Hu, who held the political whip in the early sixth century, has family origins in Jingzhou, Longdong (Gansu Sheng Wenwu Gongzuodui 1987, pp. 19–20). Many of the cave-temple sites of Longdong also show a preference for depicting the Great Departure episode and the A´soka story (Gansu Sheng Wenwu Gongzuodui 1987, p. 95).

As briefly mentioned in the above discussion of Zhuanglang Pagoda, new questions also arise. Why were these scenes selected and why are they shown independently without connections to their surrounding images? A possible answer lies in the iconization of each of the three scenes. One notices that narrative scenes rely on the law of the series, which subordinates the individual image to the whole, or partial course of the narrative. This law is followed in images depicted on one side of the Zhuanglang Pagoda's second story, where the Buddha's birth, bath, and the prophecy with astrologers are grouped together in the same pictorial space. However, the A´soka story, the Great Departure, and Parinirva¯n. a episodes are shown individually, independently, and aligned arbitrarily with images of the Buddha triad on the other sides of the same stone block. Thus, the law of series is suspended as an individual scene becomes an independent motif in isolation from the narrative cycle.

In her study on Parinirva¯n. a images, Sonya Lee points out the Parinirva¯n. a image being adapted as a Buddhist icon at Yungang, best exemplified by the separation of the Parinirva¯n. a scene from the life cycle of the Buddha in late Northern Wei. The final moment of the Buddha's demise shown in independent placement is also found in Cave 132 at Binglingsi of the early sixth century and in Cave 5 at South Xiangtangshan of the mid-sixth century (Lee 2010, pp. 45–48). Lee attributes the independent depiction of the Parinirva¯n. a scene to its inclusion into the new thematic setting of the Buddhas of Three Ages (Past, Present, and Future), which stressed the infinite, continuous presence of Buddhas. Her emphasis on the thematic setting of an image can be employed to interpret the iconization of the A´soka story and the Great Departure. The Great Departure indicates the start of the journey that leads to the prince's enlightenment. The A´soka story carries an unmistakable symbolic meaning of the future Buddhahood of the universal ruler. Despite their different narrative contexts, both scenes fit into the thematic setting about the infinity of the Buddhahood. This growing emphasis on each individual narrative further addresses an intentional selection of images to be depicted on stacked pagodas, as well as a recognition of the religious significance that can be communicated through the selected images.

#### **4. Pictorial Programs on Stone Pagodas of the Northern Dynasties**

Nevertheless, to better understand the particular emphasis on images over architectural elements on stacked pagodas from Gansu and Shanxi, it is also necessary to examine

a broader context of the production of stone pagodas at the time. This section addresses that the interest in depicting images that carry specific Buddhist teachings on stacked pagodas is a continuation of the fifth-century tradition in making single-piece pagodas (see Figure 4). The single-piece pagodas also comprise evidence for the crucial role played by pictorial programs of images in generating symbolic significance beyond architectural elements that imitate wooden buildings.

The single-piece pagodas of the fifth century are carved with Buddhist images popular at the time, including the Historical Buddha S´akyamuni, the Future Buddha Maitreya, the ¯ Fasting Buddha, narrative scenes from the Buddha's Life Story, and the Twin Buddhas motif. A comparative study of both the single pieces and the stacked pagodas showcases the exploration of how to embed meanings through the execution of images on pagodas.

One of the most representative examples of the single-piece pagodas, the Cao Tiandu Pagoda of 466, features a Buddha niche in the center on each side of the lowest story of the pagoda's main body. Clockwise, the four niches house images of S´akyamuni in ¯ dhyana mudr ¯ a (meditation gesture) on two sides, Bodhisattva Maitreya with legs crossed ¯ at his ankles, and the Twin Buddhas motif (see Figure 4) respectively.<sup>39</sup> On the remaining eight stories, each of the upper six depicts two rows of seated Buddha images on each side, while the lower two depict three rows of seated Buddha images, representing the Thousand-buddhas motif (Caswell 1975). Another three pieces showcase a similar design. One is a stone pagoda preserved in the Daiwangcheng 代 王 城 Museum, in Yu County 蔚 縣, present-day northwestern Hebei 河 北 Province (Figure 12).<sup>40</sup> The body of the pagoda has eight stories; there should have been nine stories, as pagoda usually have an odd number of stories, and the top of this pagoda is severely damaged. The bottom story is also cut in half, with the lower half and pedestal missing. Each story of the pagoda features a roof with protruding rafters and brackets, suggesting its imitation of a wooden prototype. In the center of each side of each story, there opens a niche with a seated Buddha image, flanked by smaller Buddha images in two registers on either side. The space between brackets under the roof is further carved with a small, seated Buddha image. Despite the lack of epigraphic traces that bear an exact date, the Buddha image on the Daiwangcheng Pagoda exhibits a strong stylistic resemblance to those from Phase II Yungang cave-temples of the 480s (Chen 1996).<sup>41</sup> Two other pagodas from Shanxi can be dated to the 460s and 470s. One is a three-story stone pagoda found in Datong (hereafter named the Datong Pagoda).<sup>42</sup> The other five-story pagoda was found at the Nanchan Monastery 南 禪 寺 on Mount Wutai 五 臺 in northern Shanxi province. On its second story, three sides depict seated Buddha images, while the fourth displays two Buddhas sitting side by side, representing the Twin Buddhas motif.<sup>43</sup> An approximate time range for these two works extends from the 450s to the 470s, slightly earlier or later than the Cao Tiandu Pagoda.<sup>44</sup>

Examining the alignment of the various images pictured on these stone pagodas, one finds an arrangemen<sup>t</sup> that lines up the historical Buddha S´akyamuni, Maitreya Bodhisattva ¯ as the Future Buddha, and the Twin Buddhas motif horizontally or vertically, representing the concept of the Buddhas of the Three Ages theme. The Buddhas of the Three Ages theme was already developed in Gandharan tradition, with the past Buddha commonly ¯ represented by D¯ıpa m. kara or the Seven Buddhas (including six Buddhas of the Past and S´akyamuni of the present era). Therefore, the fifth-century miniature pagodas represented ¯ the theme in a new form, which replaced the Gandharan tradition, with the Twin Buddhas ¯ motif employed to indicate the Past. The particular significance of the Twin Buddhas motif in underscoring the timelessness of Buddhahood (coexistence of the present and the past) indicates a further emphasis on the Mahay¯ ana teachings in Chinese Buddhist art. ¯ 45 In addition to the theme of the Buddhas of Three Ages, the lowest story of some of these miniature pagodas further depict scenes from the Buddha's Life Story (Ch. *fozhuan gushi* 佛 傳 故 事) in a sequence different from the sporadic depiction of similar scenes on stacked pagodas (see Figure 9).<sup>46</sup>

**Figure 12.** (**a–d**) Daiwangcheng 代王城 Pagoda, late fifth century, Northern Wei. (Yuxian 蔚县, Hebei province. Stone. H. 150 cm, W. (base) 37.5 cm. Daiwangcheng Museum. Photograph courtesy of Zhang Jianyu. Diagram by the author).

In addition to single-piece stone pagodas discovered in the Pingcheng area, those found in another cultural hub, the Hexi Corridor in northwest China, exhibit variations that correlate to Buddhist teachings circulated regionally. While retaining the horizontal scheme in depicting the Buddha's Life Story on the bottommost story, surviving miniature pagodas from the Hexi region display Maitreya Bodhisattva in combination with a Fasting Buddha image on the highest story (Figure 13). The Cao Tianhu 曹天護 Pagoda of 499 features narrative scenes from the Buddha's Life Story on its lowest story, resembling the Pingcheng tradition.<sup>47</sup> However, on its topmost story, two sides feature a seated Buddha image within a horseshoe-shaped niche (Figure 13a,d); the third side depicts Maitreya (Figure 13b), which is identifiable by his crossed ankles and the trapezoidal arch above; and the fourth side renders a skeleton-like Buddha sitting beneath twin trees, which is unmistakably the image of a Fasting Buddha (Figure 13c).<sup>48</sup> The combination of images of Maitreya and the Fasting Buddha is quite unique, without any similar specimen found in Indian and Central Asian traditions. However, a further inquiry of both textual and visual sources regarding the fasting image reveals that the significance of the combination of fasting Buddha image with Maitreya lies in its representation of the Buddha's enlightenment since its inception in Gandhara. ¯ 49 Given the Fasting Buddha's Gandhara lineage, and Maitreya's ¯ reference to future Buddhahood, the new configuration of the two on the topmost story of miniature pagoda attests a new emphasis on the attainment of enlightenment in fifthcentury Buddhist practice in the Hexi Corridor.

**Figure 13.** (**a–d**) Cao Tianhu 曹天護 Pagoda. Jiuquan, Gansu Province. 499, Northern Wei. (Stone. H. 38 cm, W. (base) 16 cm. Jiuquan Municipal Museum. Diagram by the author based on photographs in Chen 1986, Figures 1–4).

To conclude, the examination of stone pagodas made earlier than stacked pagodas in northern China shows a continuous tradition of depicting images on their surfaces. The images are imbued with Buddhist teachings popular at the time, echoing the development of the image worship found on other media, such as steles and statues. That being said, Buddhist images adorning the surface of stone pagodas have been central to the display of meaning since early on.

#### **5. Individual Image, Collective Patronage**

Despite the continuation of generating significance with images on pagodas through the fifth and the sixth century, on stacked pagodas of the sixth century we see a dissolution of pictorial programs that once organize every image adorning the surface of fifth-century single-piece pagodas, as well as a growing complexity and independence in individual

images. As examined in the third section, each image depicted on stacked pagodas is not particularly related in content with any adjacent ones. Factors shaping this growing independence of image include the trend of iconization of narrative scenes from the Buddha's Life Story and the growing importance of particular significance carried by individual narrative episodes. Nevertheless, based on surviving inscriptions at Nannieshui, we find that a stacked pagoda was often commissioned by a group of patrons. Usually, a single patron usually donates an individual image rather than a stone block or a stacked pagoda. In other words, patrons donated in unit of the individual image. To understand this specific way of dedicating stacked pagodas, this section examines the correlation between the production of stacked pagodas and the organization of patronage based on surviving epigraphical evidence. I contend that the very fact that these stacked pagodas being assembled by individual stone blocks further allows or responds to the way how they were commissioned by patrons.

The eastern Gansu and the Nannieshui material exhibit differences in terms of patronage. The eastern Gansu stacked pagodas were usually commissioned in whole by local donors. In several examples one also finds depictions of the deceased family members of the donor. For instance, the surviving character of 亡, which refers to "the deceased", on the 518 pagoda, suggests that the work was dedicated to gaining merits for the deceased family members.<sup>50</sup> Another stacked pagoda dated to 536 bears the images and inscriptions of a number of donors on the pedestal.<sup>51</sup>

On the Zhuanglang Pagoda, a procession scene of donors occupies the lower two of the three horizontal registers on one side of the bottom block (see Figure 9b). The upper register depicts a preaching scene that features a central seated Buddha. On the lower two registers, three male figures are pictured on each level riding on horses toward one direction, with cartouches indicating their identities as the donor's deceased family members.<sup>52</sup> The procession scene depicting horses and carts is commonly found on sixthcentury steles from Shanxi, Shaanxi, and eastern Gansu.<sup>53</sup> Donor figures began to be carved in an arrayed manner along the bottom of the statue and stele pedestals in the late fifth century, usually standing or kneeling towards the center. In the sixth century, donor figures usually appear in a standing position, sometimes within a frame.<sup>54</sup> A similar representation of horses and carts is found on several other pieces from the same region.<sup>55</sup> The tradition continued to be found on steles from eastern Gansu throughout the sixth century.<sup>56</sup>

The rich corpus of donor images on the Zhuanglang Pagoda may relate to the fact that its donors are from the same family clan. Such a feature of organizing patronage exclusively by family ties is prevailing in works of eastern Gansu of the sixth century. It distinguishes from the contemporaneous practice in other regions, where the *yiyi*邑邑 society became the dominant form of patronage. By the early sixth century in northern China, the collective patronage of Buddhist sculptures began to flourish, and it was usually organized through the *yiyi* society (Michihata 1967; Hou 1998, 2005, 2007; Lingley 2006, 2010). Referring to a form of Buddhist socio-religious organizations, *yiyi* is usually organized by lay Buddhists living in the same village in rural areas of northern China under the leadership of at least one monk or nun (called *yishi* 邑師), to fulfill the activities of building Buddhist steles or temples, copying or chanting Buddhist sutras for mass circulation, or holding ¯ religious rituals together.<sup>57</sup> On statues from eastern Gansu, the *yiyi* society is not seen in inscriptions until the Western Wei, suggesting a relatively later absorption of the new collective patronage form in the region.

In comparison to the eastern Gansu tradition, stacked pagodas from Nannieshui intriguingly represent a new form of collective patronage, which allows each individual donor to claim ownership over one image or a stone block. First, we do not find any Nannieshui stone blocks that are inscribed with lengthy dedicatory texts that are usually common for single-piece pagodas as well as stacked pagodas from eastern Gansu. Instead, inscriptions on Nannieshui pagodas are located around the edge of each niche, comprising only several characters in a very short length showing the donor's name and title. The surnames of the patrons sugges<sup>t</sup> that they belong to different family clans. However, little

historical documentation exists to identify more information on donors who have their names inscribed on the Nannieshui group.

Second, a stone block from Nannieshui usually features on each the four sides a separate inscription dedicated to different donors. An individual donor usually claims just one side of a stone block rather than an entire block or a pagoda. For instance, scenes no. 233, no. 234, and no. 236 on the same stone block are respectively accompanied by three patrons, Gao Wen 高 文, Wang Daoqu 王 道 渠, and Li Xiao 李 小. Each side of a stone block can be commissioned by more than one donor. Scene no. 673 is accompanied by the names of Li Hanren 李 韓 人 and Li Andu 李 安 都. Meanwhile, one person can be the donor of multiple scenes. Both scenes no. 174 and no. 175 are commissioned by Wang Niusheng 王 牛 生. The carving of the same donor's name twice on two scenes under his patronage still pinpoints the mechanism of patronage at Nannieshui; no matter how many donors are involved, the unit of donation is an individual scene out of the four sides on a stone block. This arrangemen<sup>t</sup> of inscriptions further speaks to the argumen<sup>t</sup> that the commissioning and making process of each stone block that constitutes stacked pagodas shows an image-centered system.

Although there is no sufficient written material regarding the local community at Nannieshui, the practice of inscribing a donor name by a specific image is not rare in contemporaneous Buddhist statues and steles that feature a complex pictorial program. Sixth-century patrons of Buddhist steles would inscribe the work with their own names or those of deceased family members in cartouches located right by a Buddhis niche or image. For instance, a mid-fifth-century stele discovered not far from Nannieshui features nearly two dozens of short donor cartouches directly below pictorial niches in addition to names of hundreds of donors that are listed on the bottom register on the stele (Lee 2010, chp. 1). The stele is also renowned for two niches picturing the Parinirva¯n. a scene in the identical way to the Nannieshui tradition (see Figure 11).

In general, such a practice is understood on two levels. First, the inscription of donors' names and social affiliations by pictorial niches provides a chance for them to join the same league as those Buddhist deities. As addressed by Sonya Lee, with their names inscribed side by side to pictorial space, the patrons gained access to a sacred realm where they could communicate with Buddhist deities as well as their local society. Second, a correlation is constructed between these donors and the local community, in view of the way in which steles and statues were used after production—to be displayed in public space, most likely in monastic complexes or at the intersection of routes. The stacked pagodas thus become meaningful by providing the devotees with a platform, a space, to demonstrate their religious piety to a larger group of audience.

#### **6. Dissolving the Structure: From Multilevel Pagoda to Stone Image**

The image-centered making mechanism of stacked pagodas, as address above, further sheds light on underexplored issues regarding their perception and veneration among the local community. In previous studies, there is another strand of scholarship to define the stacked pagodas under discussion, which considers each individual stone block as stele, *simianxiang* 四 面 像 (four-sided image), or *zaoxiangshi* 造 像 石 (image stone), instead of pagodas.<sup>58</sup> The reference to these pagodas as *simianxiang* or *zaoxiangshi* is primarily a created appellation based on scholarly interpretation of the depiction of images on four sides of a stone block. Nevertheless, the discussion on the definition of stacked pagodas also echoes some inscriptions found on stone blocks, which intriguingly refer to themselves as *shixiang* 石 像 or *xiang* 像.

Why are some pieces of stacked pagodas denoted as *xiang* 像 instead of *futu* or *ta*? What is the difference between *xiang* and *futu* in the context of making stacked pagodas? How do these different denotations relate to the perception and veneration of stacked pagodas? This section contends that the self-reference of *xiang* in inscriptions found on stacked pagodas is a later phenomenon that has not taken place until the later sixth century, replacing the earlier self-reference, *futu*. The primary factor that contributes to this shift

of denotation, lies in the specific emphasis on individual images or stone blocks, or *xiang*, which was fundamentally shaped by very making and commissioning process of stacked pagodas. Meanwhile, this shift also echoes with the broader historical context of the second half of the sixth century, when the growing popularity of the hemispherical stupa form ¯ posed challenges to the dominance of the multilevel pagoda form. In addition, the growing understanding of the miraculous deeds of *xiang* might also contribute to the transformation from *futu* to *xiang* among stacked pagodas.

As discussed in the introduction, several stacked pagodas bear inscriptions that refer to themselves as *shi futu san jie* 石 佛 圖 三 劫 (three-story stone pagoda), or *shi futu san ji* 石 浮 圖 三 级 (three-story stone pagoda).<sup>59</sup> Both *sanjie* 三 劫 and *sanji* 三 级 mean "three-story." The emphasis of the three levels in the inscriptions is almost universally employed in contemporaneous epigraphs that describe the construction of pagodas, therefore revealing these stacked pagodas' formal resemblance to pagoda buildings. For instance, the inscription on the Quan Pagoda of 536 also records *sanjie shi yiqu* 三 劫 石 一 区 (a three-story stone).<sup>60</sup> The inscription of the Huisuisi 暉 福 寺 stele of Duke Dangchang 宕 昌 公 reads, "I commissioned two three-storied *futu* for the two emperors at my old houses, one in the south and another in the north in my hometown."<sup>61</sup> The inscription on the construction of *futu* by Chang Huan 常 煥 and the others reads, "There is a five-story *futu* inside the temple yard."<sup>62</sup> The multilevel verticality of these stacked pagodas, as implied by their inscriptions, further reinforces the perception of them as pagodas. Wei-Cheng Lin, in his recent article on the vertical rise of Chinese pagodas in the Middle Period (10th–14th century) contends that the importance of the verticality of pagodas derives from their performative aim in drawing the attention of the faithful performative function (Lin 2016).

However, a few other inscriptions found on stacked pagodas feature the term *shixiang* 石 像 (stone image) to refer to stone blocks that comprise these stacked pagodas. The Xiejiamiao Pagoda of 558 records *shixiang yiqu* 石 像 一 區 (a stone image). Inscriptions located on the side marked as no. 392 of the Nannieshui pagoda of 553 mention the construction of *wu shixiang* 五 石 像 (five stone images), which might refer to five stone blocks including the one bearing the inscription. In addition to *shixiang*, we also find *zaoxiang* 造 像 in some inscriptions. Yet a distinction between the two terms is rarely spotted. *Zaoxiang* is found frequently in inscriptions found on contemporaneous steles, highlighting the very act of commissioning an image. The term *shixiang*, distinctively, denotes the entirety of the three-dimensional sculpture.<sup>63</sup> In the case of the Xiejiamiao pagoda of 558, we know that *shixiang* (stone image) in its inscription refers to the sculptural entity, also according to the term *yiqu* 一 區 (a piece). Inscriptions on the Nannieshui no. 392 emphasize its entirety by adding *wu* 五 (five) before *shixiang*, showing the number in total.

Comparing all surviving inscriptions, one finds that *futu* is found among the earliest pieces dated to the early sixth century, while *shixiang* is only spotted among those dated to the third quarter. This periodical gap reveals that the *futu* and *shixiang* might not be interchangeable, but successive. That being said, inscriptions reveal a transition of selfreferral terms to these stacked pagodas from "three-story stone pagodas" to "five stone images" in the third quarter of the sixth century, although their stacked forms retain the same. The new term, "stone image", pinpoints a shift in perception of these stone sculptures from pagodas to sculptural images.

How did this transition happen? An important factor lies in the way these pagodas were made with a series of stone blocks that are fully adorned with images, or the imageoriented system of commissioning stacked pagodas. In other words, the very making of stacked pagodas with richly adorned stone blocks gradually shaped, altered, and transformed the way they were perceived among the worshippers. From *futu* to *xiang*, the veneration foci shifted from the entire pagoda to individual images. From "three-story" to "five," the emphasis on the vertical rise is replaced by the number of individual stone blocks. The purpose of stacking these stone blocks to erect pagodas was gradually fading away through the decades of the sixth century. The mechanism of making pagodas from individual stone blocks, instead, seized the decisive power of defining its ontological status.

The dissolution of the vertical rise of stacked pagodas might also echo a development which took place from the mid-sixth century—one-story stupa images with the ¯ archaic hemispherical dome began to flourish, primarily in the territory of Eastern Wei and Northern Qi and the south.<sup>64</sup> Although the stupa image never achieved currency in ¯ eastern Gansu or Shanxi, the flourish of the domed stupa image from the mid-fifth century ¯ nevertheless indicates a shifted perception of the form and meaning of pagodas.

Meanwhile, another trend involved is the growing prevalence of image-making on free-standing stone surface through the fifth and the sixth century, and the awareness of image-making. *Xiang* 像, or image, became more recognized as the primary means of inscribing Buddhist teachings, displaying merits for patrons, and generating religious meanings. The growing importance of *xiang*, or images in the sixth century, on the micro level, is already pinpointed above by the emphasis, highlight, and iconization of individual images that carry specific meanings, such as the Parinirva¯n. a scene. *Xiang*, or "image," is usually distinguished from other forms in terms of their roles primarily as objects of worship (Wong 2004, introduction). The veneration of images in China has also developed over the course of centuries. The establishment of image making or the image cult and its correlation to the dissemination of Buddhism in China since the second century has been a key issue in debate in the field of art history as well as religious studies. Minku Kim, in his dissertation on the very topic, has combed through the historiography of the topic and provided a definition of Buddhist image worship in a Chinese context (Kim 2011, 2014). Following his scholarship, the essential factor defining image worship is related to whether there is any ritualistic Buddhist significance displayed. Furthermore, recent scholarship in Buddhist studies also shows an intentional construction of the image worship as a distinctly Buddhist practice since the late fifth century (Greene 2018). It was a gradual process for the image worship to be integrated into medieval Chinese society.

## **7. Concluding Remarks**

This paper discusses the phenomenon of making stacked pagodas in sixth-century northern China. On the one hand, stacked pagodas were complementary in form and placement to real pagoda buildings. On the other hand, they feature their own peculiarity in terms of material, making, and venerating. Images on stacked pagodas have provided a wonderful outlet for the unbridled imagination of the artisan. Although there is a variety of delicate details to defy an attempt to impose a systematic classificatory scheme of arranging images on stacked pagodas, a survey of surviving images reveals a specific preference for certain images and allows further study of the context of their correlation with the broader artistic milieu. More importantly, an investigation of the relation between carved images and inscribed texts against the historical context of dedicating Buddhist statues and constructing pagodas sheds light on the making and perception of pagodas in non-built form in sixth century. In addition to the definition of sacred space with architectural forms, this case study pinpoints the quintessential role played by the adornment of images in generating sacredness. **Funding:** This research received no external funding. The APC was

covered by research funding from the Center for Global Asia, New York University Shanghai.

**Conflicts of Interest:** The authors declare no conflict of interest.
