Reconstructing the Archaeological Context of Free-Standing Buddhist Images: Considerations of the Wanfosi Hoard in Chengdu (Sichuan)

Abstract

Unlike a modern archaeological excavation of a Buddhist hoard, this article focuses on a collection of free-standing Buddhist images retrieved from at least four unofficial and intermittent excavations between 1882 and the 1950s. The excavation site in question is associated with the former site of Wanfosi, a historical monastery that has been destroyed and rebuilt multiple times from the early Tang to late Qing. By examining various sources, including antiquarian scholarships, local gazetteers, pre-modern maps, and important Buddhist images found within the hoard, I aim to provide essential information that pertains to modern archaeology. This article seeks to address issues such as the accurate identification of the excavation location, the provenance of all materials from a single Buddhist hoard, the historical lineage of the monastery from which insights are gathered into the image cache, and the noteworthy observation that the preservation of the pre-Tang images occurred long before the final burial.

1. Introduction

Since the late nineteenth century, if not earlier, a collection of Buddhist images has been retrieved from an urban site in Chengdu 成都 (Sichuan). The site in question has been associated with the ruins of Wanfosi 萬佛寺 (“Monastery of Ten Thousand Buddhas”) that might once have been in-place according to some late historical information. Unlike a modern archaeological excavation, the discovery of the hoard of Wanfosi was not a one-off excavation. Instead, unofficial excavations took place at the former site of Wanfosi in 1882, followed by 1937, 1945/1946, and 1953/1954. It was no other but the great erudite antiquarian Wang Yirong 王懿榮 (1845–1900) who documented the first excavation event of 1882 in his Tianrangge zaji 天壤閣雜記 (Miscellaneous Reports on the Heaven and Earth Pavilion, hereafter abbr. as TRGZJ),1 a short volume of selected antiquarian records compiled no later than 1885.2 The following discoveries between 1937 and 1954 were disclosed in two critical works by modern scholars in the 1950s: an article titled “Chengdu Wanfosi shike zaoxiang” 成都萬佛寺石刻造像 (“Stone Carvings and Images of Wanfosi, Chengdu”, hereafter abbr. as CWSZ) by Feng Hanji 馮漢驥 published in 1954 (Feng 1954), and the first catalog titled Chengdu Wanfosi shike yishu 成都萬佛寺石刻藝術 (Art of the Stone Carvings of Wanfosi, Chengdu, hereafter abbr. as CWSY) by Liu Zhiyuan 劉志遠 and Liu Tingbi 劉廷壁 published in 1958 (Liu and Liu 1958). Subsequent to several discoveries that resulted from mass-scale construction projects across urban Chengdu in the following decades, the provincial museum and other local relevant institutions published a comprehensive catalogue titled Sichuan chutu Nanchao shike zaoxiang 四川出土南朝石刻造像 (Buddhist Statues of the Southern Dynasties Excavated in Sichuan, hereafter abbr. as SCNSZ) in 2013 (Sichuan Bowuyuan et al. 2013).

The Wanfosi hoard, possibly buried in late Tang 唐 (618–907) or even later, surprisingly consists of free-standing images almost exclusively dating from the Liang 梁 Dynasty (502–557). These finds were particularly significant for Chinese medieval Buddhist art, as scholars had long been aware of the rarity of Buddhist archaeological remains from the south, especially in comparison to their Northern counterparts. Even before the first discovery from Wanfosi, epigraphical works of the Qing 清 Dynasty (1644–1911) had already noted the inaccessibility of the Southern materials. In his epigraphic work, Yushi 語石 (On stones), Ye Changchi 葉昌熾 (1849–1917) wrote of the forging of Southern Dynasties’ Buddhist artworks during the Daoguang 道光 (1821–1850) and Tongzhi 同治 (1862–1875) Eras, as a result of the rising demand for Southern materials in the thriving antique market in later phases of Qing (Ye and Ke 1994, pp. 312, 328). The rarity of the Southern materials remains the same today. Although new discoveries of stone images were made in urban Chengdu after the 1990s (Zhang and Lei 2001; Chengdushi Wenwu Kaogu Gongzuodui and Chengdushi Wenwu Kaogu Yanjiusuo 1998; Chengdushi Wenwu Kaogu Yanjiuyuan 2017), and sporadic bronze images were reported from Nanjing 南京 (Jiangsu) in recent decades (Fei 2018), the irreplaceable importance of the Wanfosi images in question is endorsed by their fair quantity, massive size, abundant iconographies, and skilled craftsmanship.

Because of the unparalleled importance of the Liang materials from Wanfosi, in the early stage, scholars researched several critical topics. For instance, several images from the site, especially those with questionable iconographies, aroused heated discussion among scholars. Alexander C. Soper (Soper 1960), Yang Hong 楊泓 (Yang 1964), Nagahiro Toshio 長広敏雄 (Nagahiro 1969), and Yoshimura Rei 吉村怜 (Yoshimura 1985) made diverging suggestions in the early stage of research on the Wanfosi finds. In the following decades, the interpretation of several stele-shaped images still aroused intense debate (Wong 1998; Zhang 2004; Li 2010), while several Chinese academic studies were dedicated to providing a comprehensive overview of the Southern Dynasties (420–589) images found in Sichuan, in which the Wanfosi materials play a crucial role (Li 2000; Lei 2009, 2018). The works by Dong Huafeng 董華鋒 and He Xianhong 何先紅 published in 2014 (Dong and He 2014; He and Dong 2014) are the most relevant for the present study that investigates the Wanfosi hoard per se, because they review the multiple excavation campaigns and the materials related to Wanfosi.3 However, unlike a modern archaeological excavation of a Buddhist hoard, the so-called Wanfosi hoard is a collection of Buddhist images from at least four unofficial and intermittent excavations with very limited written information in the literature, and there seem to be more undocumented digs that occurred.4 Excavated long before knowledge of modern archaeological methodologies reached China in the 20th century, most Wanfosi materials lack a record, which present-day archaeologists fundamentally give close attention, including the following information: precise location, stratigraphy, excavation units, etc. As a result, this situation led to several research questions focused on in this article. First, according to the publications currently available, the statements regarding the location of Wanfosi are inconsistent. Being neglected by the previous scholarship, it is my own observation that the pre-modern sources and the publications in the 1950s narrow down to one area or location, while, unexpectedly, the comprehensive catalogue SCNSZ in 2013 pinpoints to another location more than one kilometer to the northeast. This deviation could have led to the following basic question: whether it is one integrate hoard or two different hoards instead. In Section 2, upon carefully reviewing the extensive sources of the four unsystematic excavations, I propose that the dispute over the location might have been caused by a relocation event that is recorded in TRGZJ. Second, the excavated Buddhist images have been ascribed to “Wanfosi” in general, without considering the vicissitudes of this institution, which was founded, fell into ruins, experienced multiple phases of revival, and finally perished in the 1930s. In Section 3, I investigate the fundamental problems that arise from such an identification, in order to better perceive the timeline of the formation, burial, and discovery of the Buddhist hoard. In addition, it is necessary to confirm if the hoard was ever opened prior to 1882. In fact, TRGZJ once associated the excavated stone images with the Zhang Xianzhong’s 張獻忠 (1606–1647) atrocity, but I consider it as a misquotation from a previous mid-Qing work. Upon a clearer understanding of the modern archaeological context and the historical lineage of Wanfosi, Section 4 probes into the formation of the hoard: that is, how the Liang images were assembled and eventually buried as a “hoard”. Based on an estimated date of burial and chronology of the hoard, a particular focus is given to a special stone image, which had gone unnoticed by previous scholars: Produced during the early period, it was reused at a much later time. I argue that this special image testifies to the practice of preserving earlier Buddhist images during later phases of the monastery, a practice that provides some fresh insight into the formation of the Buddhist hoard.

2. Precise Location of the Hoard

The exact location (or locations) is (or are) not absolutely clear to modern scholars. As a matter of fact, different statements were given concerning the exact present-day location of Wanfosi. In addition, because the excavation was not a one-off event but multiple campaigns across seven decades, it is also necessary to determine whether several of these campaigns took place in the same geographic location. These ambiguities originated from the lack of a modern archaeological context lead to the fundamental question in this section: do these Buddhist images originate from a single hoard, or different hoards instead?

2.1. Records from Pre-Modern Sources

In Wang Yirong’s TRGZJ, he first provenanced the Buddhist images discovered in the late nineteenth century with an elusive location: “At the West Gate of Chengdu, there was the former site of Wanfosi” (Chengdu Xiguan you Wanfosi guzhi 成都西關有萬佛寺故址).5 According to the Chongxiu Chengdu xianzhi 重修成都縣志 (Revised Prefectural Gazetteer of Chengdu) compiled in 1873, Wanfosi was closely located nearby a now-lost bridge, Jinhuaqiao 金花橋.6 The relative geographical positions of Wanfosi and Jinhuaqiao are also reflected in at least two pre-modern maps, respectively, made in 1879 and 1911 (Figure 1). Both maps show Wanfosi situated outside the now-destroyed West Gate, named Qingyuanmen 清遠門, on the riverbank of a moat that surrounds the city walls, and to the northwest of the Jinhuaqiao bridge. Qingyuanmen was recorded as the “West Gate” in TRGZJ. This gate was one of the four city gates constructed in the Qing period. Its original location should have been roughly at present-day Xiyue Cheng 西月城 Street (Qingyang 青羊 District), northwest downtown Chengdu (Figure 2, Location 1, Yuan 2017, pp. 69–70).7 The moat outside Qingyuanmen should be the present-day Xijiao 西郊 River crossing with Xiyue Cheng Street. On both pre-modern maps, Jinhuaqiao crosses the first waterway to the north of Qingyuanmen, which flows to the moat (Xijiao River). By comparison with the present-day hydrographic net, the now-lost Jinhuaqiao should be located approximately at present-day Jinxianqiao 金仙橋 Road (Jinniu 金牛 District, Figure 2, Location 2). Therefore, the putative geographical scope of Wanfosi, as recorded in TRGZJ, Qing gazetteers, and pre-modern maps, should be the area northwest of Jinhuaqiao (Figure 2, Location 3).

2.2. Clues from Modern Excavations

The next documented excavation was in 1945/1946. In terms of the location, multiple sources pointed to the erstwhile “Sichuan College of Science” (“Sichuan Lixueyuan” 四川理學院, Feng 1954, p. 110; Liu and Liu 1958, p. 3). However, its name did not indicate that it was an institution throughout the Republic period. Based on a thorough search of newspaper archives, there is no institution entitled as such. Instead, it should be Chengdu College of Science (Chengdu Lixueyuan 成都理學院), which was founded in 1945 by the mathematician Wei Shizhen 魏時珍 (1895–1992).8 The institution finished the construction of three major buildings in 1947 centred around the “Garden of the Ma Family” (Majia Huayuan 馬家花園) outside the “West Gate”.9 This private garden, built approximately in the 1940s, has long vanished, but the location should be the present-day hostel area of the China Railway No. 2 Group (Zhongtie Erju 中鐵二局, Figure 2, Location 4, Yuan 2017, pp. 1141–42). As a result, Location 4 should be the excavation location for the 1945/1946 campaign without much deviation. For the excavation in 1953/1954, the same company, which operates in the railway industry, reported new finds during construction (Sichuan Bowuyuan et al. 2013, p. 6), and the location should coincide with that of the 1945/1946 excavation (Location 4). Eventually, as shown in Figure 2, the location of the excavations in 1945/1946 and 1953/1954 was highly consistent with the putative geographical scope of the excavation in 1882, as documented in TRGZJ.

It was not until 2015 that an archaeological excavation, conducted at locus 13 of Tongjin 通錦 Road in Jinniu District (Figure 2, Location 5), shed new light on the Wanfosi site (Chengdu Wenwu Kaogu Yanjiuyuan 2018). The excavation of Tongjin Road yielded two sandstone-made funerary land deeds (maidi quan 買地券), dating from the Tang and the Song inscribed with characters such as “jingzhong” 淨眾 (“pure assembly”) or “jing” 淨 (“pure”) (ibid., pp. 182–84). Therefore, this site is considered to be the garden area of the Jingzhongsi of Tang, that is, the putative former site of Wanfosi, the historical lineage of which is discussed in Section 3 (ibid., pp. 204–5). Therefore, the connection between Tang Jingzhongsi and Qing Wanfosi may be well reinforced once again. If Locations 3 and 4, as shown in Figure 2, are indeed the precise location of the Wanfosi site, the location of the Jingzhongsi Garden is adjointed just to the east, which testifies to the connection between the materials that were excavated at Tongjin Road and the historical lineage of the Wanfosi.

2.3. A Different Statement on the Wanfosi Location

In contrast to the previous narrative, however, a critically different statement on the location of Wanfosi can be found in the comprehensive catalogue SCNSZ. The authors contend that the ruins of Wanfosi should be located in northwest Wudingqiao 五丁橋 (connecting Jinniu and Qingyang Districts), approximately at the intersection of Baimasi 白馬寺 Street and the North 2nd Section of the 1st Ring Road (Figure 2, Location 6, Sichuan Bowuyuan et al. 2013, p. 5). To endorse such an identification, SCNSZ also sought out testimonies from local elderly residents, who are supposed to have some in-person experience with the excavation campaigns (ibid.). However, as referenced by the aforementioned pre-modern gazetteers and maps, Wanfosi was located “northwest of the town”, and it should be northwest of Jinhuaqiao. However, the location identified in SCNSZ is north of the town and to the northeast of Jinhuaqiao. In addition, it is more than one kilometer from Locations 4 to 6 (Figure 2), which is too distant to consider Location 6 as part the Wanfosi ground plan. Moreover, the Fu 府 River separates both sides to the east and west, which also make it challenging to visualize the architectural layout of the monastery, if it ever existed. Obviously, the location identified by SCNSZ does not belong to the same ruins of Wanfosi (Figure 2, Location 4). Instead, it should belong to another location nearby the original one. In the above discussion, the location of the excavations in 1882, 1945/1946, and 1953/1954 were consolidated (Figure 2, Locations 3 and 4), while only the 1937 event, or any other undocumented digs, lack geographical records. Is it possible that the “local elderly residents” may have information that was handed down or even witnessed the 1937 event or any undocumented excavation attributed to the reported Wanfosi origin? There seems to exist undocumented excavations in addition to the four noted above.10 If the 1937 campaign, or any undocumented dig, was indeed conducted at Location 6 (Figure 2), there could be another scenario that involves a more complicated narrative: the possible relocation of the first batch of images as recorded in TRGZJ. The relevant passage reads as follows:

All (stones) are headless, or, preserving the heads without bodies, and not a single (stone) was intact. This was what Shubi (Elegy of Shu) claims to be chiseled away by the Xian thieves. It was reported by two counties that the obtained (stones) reached a total of more than a hundred. The great man of the family [jiadaren, also known as “my father”, Wang Zuyuan 王祖源 (1822–1886)] commanded the localities to relocate (the stones) to present Xiao Wanfosi, he financed the restoration and had them fully repaired. (He) commanded us brothers to supervise the project, without spending one penny of official funds, or one penny of (collected) donations.

皆無首或有首無身, 無一完者, 蜀碧所稱獻賊鑿去者也. 兩縣來報, 出凡百餘. 家大人命地方移送今小萬佛寺, 出資重完且盡整之. 命余兄弟監其事, 不用官家一文, 一文不募.11

In the above text, another monastery, Xiao Wanfosi (small-scaled Wanfosi), is mentioned. Xiao Wanfosi is the site at which the Wanfosi materials were once “relocated to, restored and repaired” by the Wang brothers [Wang Yirong and Wang Yiqi 王懿棨 (1851–1899)]. Judging from the monastery’s name, there must have been a certain connection between Wanfosi and Xiao Wanfosi. By 1882, the historical Wanfosi had already been demolished. Supposedly, this new monastery was not too remote from its archetype, or it would not have been named after the original Wanfosi. Therefore, there is the possibility that the Xiao Wanfosi mentioned in TRGZJ was actually Location 6 (Figure 2). This possibility is consistent with the testimony of “local elderly residents” and was adopted by the catalogue SCNSZ. In addition, according to the cataloging work of the Sichuan Museum, there are a total of 12 extant life-sized images obtained from the 1937 campaign, including nine Buddha torsos and three heads (Sichuan Bowuyuan et al. 2013, p. 18). The status of these images matches the textual record in TRGZJ: all images have their heads and bodies separated (wushou huo youshou wushen). Therefore, Location 6 is possibly Xiao Wanfosi, where the stone images obtained in 1882 from the nearby Wanfosi site were “relocated to, restored and repaired” in situ, and later discovered in 1937.

2.4. Discussion of the Location

The above hypothesis of image relocation, however, is merely based on the testimony of “local elderly residents” and the description in the official catalogue SCNSZ. The identified site of Location 6 (Figure 2) is far from being treated as concrete evidence, as false testimony or errors that may occur during the process of paraphrasing or editing the catalogue may be misleading. Yet, since Location 6 is clearly stated in an official, comprehensive catalogue, and different locations may lead to multiple hoards of the Wanfosi materials, it is my intention to explore every possibility for the reconstruction of the lost archaeological context in the most logical manner. We are now able to probe the most important question: are the Wanfosi images from a single hoard or multiple hoards? On the one hand, if Location 6 is indeed the Xiao Wanfosi excavated in 1937, the obtained stone images (the 12 life-sized images currently preserved in the provincial museum) were once relocated from the Wanfosi site (Location 4), and restored and repaired by the Wang brothers in 1882 as recorded in TRGZJ. On the other hand, if Location 6 is entirely irrelevant to the Wanfosi site, having been suggested as such due to the fact of being misled by the false testimony, the 1937 excavation, or any other undocumented digs, should have also taken place at Location 4. Eventually, no matter which hypothesis represents the historical fact, the excavated materials from four different excavation campaigns, namely, the 1882, 1937, 1945/1946, and 1953/1954 campaigns, should come from a single hoard (Location 4). As previously discussed, this geographical location is within the hostel area of the present-day China Railway No. 2 Group. The latest event that led to a mass-scale discovery is reported to be infrastructure construction conducted in 1953 (Liu and Liu 1958, p. 3). Upon physical on-site investigation, the only architecture that dated to such a period is a Khrushchevka-style brick, five-storied apartment building (Figure 3). Therefore, the location of the former site of Wanfosi, where multiple batches of Buddhist images were excavated and reburied, should be underneath this building.

If we merely sum up the numbers from different textual records, the total number of excavated specimens from the four excavations could easily reach several hundred. The 1882 event reported “more than one hundred” images; the 1937 event yielded a total of 37 specimens; the 1945/1946 campaign unearthed “a satisfactory amount of Buddhist images;” and the 1953/1954 event yielded “approximately 200 specimens” (Liu and Liu 1958, p. 3). However, since all stone images presently obtained are proved with certainty to originate from a single hoard, there is the possibility that some images that were obtained later may have been reburied during previous events. For instance, the Buddhist images discovered during the 1945/1946 campaign were “smashed, destroyed, or buried underneath the groundwork” (Liu and Liu 1958, p. 3). This batch of materials could be the approximately 200 specimens that were later found during the 1953/1954 excavation at the exact same location. This relocation theory is also discussed in relation to Wanfosi and Xiao Wanfosi. This shows the possibility that the extant specimens from the 1937 campaign (or any other later undocumented discoveries) could have originated from the 1882 excavation. Moreover, as the Wanfosi materials are generally in a fragmental condition, the number of found images is usually determined on a “piece-by-piece basis”. Consequently, the number of integrated images could be much lower. Considering these factors, the estimated total number of Wanfosi materials ever to be excavated may not be as exorbitantly high as stated in the textual records.

3. Historical Lineage of the Monastery Wanfosi

In the previous sections, I followed the common practices of this field, which have been established since the 1950s, at the latest, to refer to this group of nonscientifically excavated specimens as the Wanfosi hoard, precisely because of the lack of a better term and also to avoid ambiguities. The current section, subsequently, concerns the fundamental problem with this designation, which is the historical lineage of the particular monastery in situ, which helps determine the Buddhist hoard, especially the more intriguing pre-Tang (before 618) materials. Long before the excavations took place, Wanfosi had endured as a reputed historical monastery. Wanfosi’s historical name, Jingzhongsi 淨眾寺, can be traced in the Buddhist literature, such as Song gaoseng zhuan 宋高僧傳 (Song Biographies of Eminent Monks, T2061),12 and Lidai fabao ji 歷代法寶記 (Record of the Dharma-Jewel through the Ages, T2075).13 Zhang Zikai’s 張子開 previously published study (Zhang 1999) is still the most relevant reference to study the historical lineage of the monastery. However, this article considers the Buddhist images that were preserved in the early stage of the monastery only to a small degree. Understanding the historical lineage of the monastery is crucial for gaining insights into the image hoard, which was initially preserved in the early stages, subsequently buried underground, and eventually unearthed from the ruined site. On the other hand, an event cited in TRGZJ indicates that the hoard had been opened during the early Qing period. It is thus essential to determine whether the hoard had ever been “interrupted” prior to the 1882 excavation. Therefore, this section reviews the gazetteer records to establish a clear-cut historical lineage of the monastery, investigates whether the hoard was “interrupted” in the early Qing period, and extracts clues that help to better understand the preservation, burial, and excavations of the hoard.

3.1. Ming and Qing Gazetteers

Wang Yirong’s TRGZJ is one of the earliest textual records concerning the first excavation in 1882, which refers to Wanfosi as a “former site” (guzhi).14 If the monastery was already utterly ruined by 1882, the Wang brothers, along with the scholarly community in the late Qing, should have obtained the historical information and the geographical location regarding Wanfosi from pre-existing works, especially from provincial and prefectural gazetteers. The earliest record that exists in local gazetteers appears to be the Sichuan zongzhi 四川總志 (Assembled Gazetteer of Sichuan) compiled during the Jiajing 嘉靖 Era of Ming (1522–1566):

Jingyinsi (Monastery of Pure Cause), (located) to the northwest of the prefectural city, colloquially called Wanfusi.

淨因寺, 府城西北, 俗呼萬福寺.15

This record was collected entirely in the Xinxiu Chengdu fuzhi 新修成都府志 (Revised Prefectural gazetteer of Chengdu), composed in the inaugural year of the Tianqi 天啟 Era of Ming (1621).16 Nevertheless, the Tianqi edition has recorded a lengthy “Stele-inscription on the Reconstruction of Wanfusi” (Chongjian Wanfusi beiji 重建萬福寺碑記), which is partially transcribed as below:

Jingyinsi, colloquially called Wanfosi, has recently changed its “fo” into “fu”. According to tradition, it was constructed during the Yanxi Era of Han (158–167). Some say that (Jingyinsi) is the ancient Jingzhongsi, which is located at the site of the ancient Zhulinsi. Chan Master Musang (Ch. Wuxiang, d. 762) of Tang built a stūpa and had ten thousand Buddhas sculpted, and the monastery was thus named. When later the stūpa fell into ruins, the Military Commissioner (Hucker 1985, p. 144, entry no. 777) Gao Pian (821–887) took (the bricks) from the stūpa and constructed the barbican… In the middle of the Hongwu Era (1368–1398), King Xian of Shu assumed (the post of) the kingdom, the (construction of the) palace had not been completed, (and he) frequently visited the place. The remaining images all existed, at that time (the monastery) was still called Zhulinsi. Monk Zhongxuan (J. Naka Era?) from Japan practiced Chan (meditation) there. The King of Shu was moved by his sincerity, with extra-money from (his revenue as) a Ming imperial prince,17 he made offerings of golden images, and placed the dharma treasure... In the middle of the Zhengde Era (1506–1521), the monastery was burned down by the roving bandits, only the halls remained unaffected… At the beginning of the Wanli Era (1573–1620)… It was the eighth month of the jiachen year (1694)… the repair and restoration were completed… At first, the King of Shu expended the state property, and ordered laborers to repair… which costed as much as three or four thousand maces of gold.

淨因寺, 俗呼萬佛寺, 近又易佛為福矣. 相傳創於漢延熹. 或曰, 即古淨眾寺. 古18竹林寺地. 唐無相禪師建塔鐫佛者萬, 寺以故名. 後塔毀, 高節度使駢取修羅城… 洪武中蜀獻王就國, 宮未就竣, 多游其地, 遺像具在, 時猶號竹林寺. 日本僧中選者, 禪焉. 蜀王感其誠, 以承運殿副材, 施奉金像, 並置法藏… 正德中, 寺燹於流賊, 惟殿無恙… 萬曆初…歲甲辰八月… 修復之完… 先是, 蜀王出帑金, 命工修葺… 費可三四千金.19

The Sichuan tongzhi 四川通志 (Provincial Gazetteer of Sichuan) was compiled in the eleventh year of the Yongzheng 雍正 Era (1733), and was published with the first brief entry on Jingyinsi as seen in the Jiajing edition, but the book does not provide new knowledge concerning the monastery.20 In the next gazetteer edition, in the Chengdu xianzhi 成都縣志 (Prefectural Gazetteer of Chengdu) compiled in the 21st year of the Jiaqing 嘉慶 Era (1816), there is a passage related to Jingzhongsi containing additional information:

Jingzhongsi, (located) in the northwest of the county (town)… was named Jingyinsi in Song, and was renamed Wanfosi in Ming. There was an enormous bell, weighing a thousand jun, which is currently abolished.

淨眾寺, 在縣西北… 宋名淨因寺, 明改為萬佛寺. 有巨21鐘, 重千鈞, 今廢.22

Complied in the same year, 1816, the Sichuan tongzhi 四川通志, Jiaqing edition (Provincial Gazetteer of Sichuan) also quotes the above passage.23 Subsequently, in the Tongzhi 同治 Era, a new edition entitled Chongxiu Chengdu xianzhi, which is also mentioned in Section 2, was the most recent publication shortly before the excavation. In terms of Wanfosi, the relevant passage reads as follows:

Wanfosi, is located near Jinhua Bridge about one li from the sixth district to the west of the county town. Gaoseng Zhuan (records that), Monk Musang, who is from the Kingdom of Silla, in the sixteenth year of the Kaiyuan Era (728), arrived Chengdu. (Musang) collected alms from patrons, and constructed Jingzhongsi (Monastery of Pure Assembly). The Hall of Images existed therein. Later, (the monastery) that was formerly named Jingyinsi was renamed Wanfusi (Monastery of Ten-thousand Blessing) at the end of Yuan or beginning of the Ming. It was ruined by Xian, the heister (Zhang Xianzhong), in late Chongzhen Era. In the (current) dynasty of the nation, it was repaired during the early years of the Kangxi Era (1662–1722), and was renamed Wanfosi. In the fifty-third year of the Kangxi Era (1714), the main hall was built. There was an ancient bell of the Tang period, which was moved and placed in the Drum Tower during the years of the Yongzheng Era. For details, see Jinshizhi… Mingshengzhi (records that), Jingzhongsi has one enormous bell weighing one thousand jun. In the Huichang Era of Tang (840–846), it was destroyed without exception. The bell was thus moved into Taicisi (Monastery of the Great Mercy). In Dazhong Era, (it was) again returned.

萬佛寺, 縣西六甲里許金花橋側. 高僧傳, 僧無相, 新羅國人, 唐開元十六年至成都, 募化檀越, 造淨眾寺, 影堂在焉… 後故名淨因寺, 元末明初更名為萬福寺, 崇禎末毀於獻賊. 國朝康熙初年重修, 改為萬佛寺. 康熙五十三年建大殿, 唐時古鐘一口, 雍正年間岳鐘琪移置鼓樓, 詳見金石志… 名勝志, 淨眾寺有一巨鐘, 重千鈞. 唐會昌例毀, 此鐘乃移入太慈寺, 大中復還.24

3.2. A Misleading Quotation from Shubi

As recorded in the Tongzhi edition of the gazetteer (Revised Prefectural Gazetteer of Chengdu), Ming Wanfosi (back then, named Wanfusi) was ruined by the Zhang Xianzhong force. Zhang Xianzhong established the Daxi 大西 Regime and ascended the throne in 1644. In the following years, until his death in 1647, multiple Qing sources condemned the tragic massacres in Sichuan launched by Zhang’s forces.25 The passage cited above from the Tongzhi edition of the gazetteer records the demolition of the monastery’s aboveground architectural structure. Still, there is no indication that the buried stone images had already been exposed by then. However, by further looking into the passage in TRGZJ, the excavated stone images were seemingly also associated with Zhang’s atrocities, judging from a quote in Shubi, a chronicle by Peng Zunsi 彭遵泗 (jinshi 進士 1737), composed in 1745.26 After a commentary on the fragmented condition of the stone images (“not a single stone was intact”), the quotation claims that they had been “chiseled away by Xian thieves” (Shubi suocheng xianzei zaoquzhe ye).27 If the quotation from Shubi in TRGZJ is valid, the Wanfosi materials would have already been exposed as early as the early Qing, and even destroyed (thus, reburied) during the Zhang’s atrocities in approximately 1645. This is rather unexpected because there is no other evidence pointing to an earlier excavation of the Wanfosi site that predates 1882. However, upon careful inspection of Shubi, there is no such record of Zhang’s forces demolishing stone Buddhist images. There is, however, only one passage that mentions Zhang’s treatment of the Buddhist images, as follows:

At that time, the (Xian) heisters established the Bureau of Casting, took the ancient ding-pots and entertaining utensils stored by the frontier office (Hucker 1985, p. 207, entry no. 1868), as well as the bronze images from the monasteries inside and outside the town, and melted them (into) liquid for (casting) cash coins, the characters on the coins read “Dashun tongbao”… All heads of the sculpted divinities did not change, although they were forged a hundred times. In the end, the thieves discarded them. Later, the Prefect of Chengdu of the current dynasty Ji Yingxiong (juren 舉人 1642) collected and buried them outside the north gate, the title of his tablet reads “tomb of the Buddha”.

是時賊設鑄局, 取藩府所蓄古鼎玩器, 及城內外寺院銅像, 熔液為錢, 其文曰“大順通寶”… 諸神像首, 百煉不化, 賊盡棄之. 後本朝成都知府冀應熊拾而埋之北關外, 題其碣曰 “佛塚”.28

This passage, which is similar to Wang Yirong’s quotation, also mentions the “heisters” (or “thieves”, zei) and “image heads” (xiangshou). However, Shubi merely documents the disposal of bronze Buddhist images instead of stone images. In addition, the final burial location, if reliable, is outside of the north gate; therefore, it does not coincide with Wanfosi’s location, which has been recorded as being located “outside of the West Gate”. It is, thus, possible that Wang mistakenly cited this passage from Shubi and paired it with the fragmented Wanfosi stone images. Therefore, the connection between the Zhang’s forces and the excavated Wanfosi materials is unreliable. There is still no evidence showing any disclosure or treatment of the Wanfosi stone images prior to the excavation in 1882.

3.3. Summarizing the Historical Lineage of Wanfosi

Based on the textual records described above, one can obtain the brief successive development of this historical monastery as illustrated in Figure 4. First, it was possibly built by, or at least closely associated with Monk Musang, the Chan master of Tang (ca. 728). During the Tang period, it was named “Jingzhongsi”. This name was used until a much later period, as the Jiaqing edition (1816) gazetteer described the monastery as “Jingzhongsi”. During the Huichang 會昌 Persecution (840–846), Jingzhongsi was destroyed but later recovered during the Dazhong Era (847–860). In the Song period, the monastery was renamed “Jingyinsi”, a term that was also continuously used, along with its successive names, until at least the Tianqi Era, since the Tianqi gazetteer (1621) introduced the monastery using the term “Jingyinsi”.29 Around late Yuan and early Ming, it was renamed “Wanfusi”, but the phonetically similar “Wanfosi” should be circulated as a colloquial name for no later than this period, and possibly even much earlier, since the Monk Musang of Tang “built a stūpa and had ten thousand Buddhas sculpted”. During the middle phase of the Hongwu Era (ca. 1389), the King Xian of Shu (Zhu Chun 朱椿, 1371–1423) sponsored the renovation and made offerings to the monastery. In the middle of the Zhengde Era (1506–1521), the monastery was burned down by the roving bandits, and only the hall remained unaffected. At the beginning of the Wanli Era (1573–1620, especially 1604), the Chan monks renovated the halls, sponsored by the King of Shu (possibly Zhu Xuanqi 朱宣圻, d. 1612, r. 1561–1612). Between the late period of the Chongzhen Era and the early Shunzhi Era (ca. 1644–1646), the monastery was destroyed during the atrocities committed by Zhang Xianzhong’s forces. During these atrocities, albeit TRGZJ recorded the demolition of stone images, the underground buried materials should not have been exposed by then. Subsequently, it was rebuilt in the early period of the Kangxi Era (r. 1662) and named “Wanfosi” afterward, and its main hall was built in 1714. Thereafter, there are no further gazetteer records related to the monastery’s evolution after the Kangxi Era (see Figure 4).

3.4. Clues from the Historical Lineage

Before delving into Section 4, which focuses on the formation and burial of the hoard, several clues can be initially gleaned from examining the evolution of the monastery. The earliest-dated Tang image from the hoard dated 728 (

Table 1. Dated images excavated from Wanfosi.

No.	Dating	Dynasty	Brief Description	Type
1	523	Liang	Śākyamuni image made by Kang Sheng 康勝	Group-configured
2	525	Liang	Śākyamuni image (devotee’s name unrecognizable)	Group-configured
3	529	Liang	Śākyamuni image made by old lady Jimo 籍莫	Standing Buddha
4	533	Liang	Śākyamuni image made by Shangguan Faguang 上官法光	Group-configured
5	537	Liang	Image made by Hou Lang 侯朗	Standing Buddha
6	548	Liang	Avalokiteśvara image made by Monk Fa’ai 法愛	Group-configured
7	562–565	Northern Zhou	Aśokan image made by Yuwen Zhao 宇文招	Standing Buddha
8	567	Northern Zhou	(No other extant inscription except for the yearmark)	Seated bodhisattva
9	728	Tang	“Stone image”	Unknown
10	737	Tang	Bodhisattva image	Unknown
11	815	Tang	Śākyamuni image re-inscribed with a Tang inscription	Group-configured
12	847	Tang	dhāraṇī pillar made by Huiguang and others	Dhāraṇī pillar
13	581–600	Sui	Textual record in TRGZJ	Unknown

), the exact date when Monk Musang collected alms and had Jingzhongsi constructed. He also had “ten thousand Buddhas sculpted”, which reflects the production of new images, doubtlessly including the Tang images of the hoard, began to flourish since the Era of Musang. As there are four Tang images dated between 728 and 847 (Table 1), the production and veneration of Buddhist images should continue to thrive until the Huichang Persecution. The stūpa constructed by Musang, which was ruined during the Huichang Persecution, possibly has not undergone any repairs during the Dazhong restoration (847–860), because Gao Pian, the Military Commissioner of Jiannan–Xichuan 劍南西川, collected the bricks from the stūpa and constructed the barbican no later than 876.30 Therefore, although the place underwent a general restoration shortly after the persecution, certain parts of the monastery have not been restored to their original condition. Along with the stūpa, it is highly possible that the image cache was discarded in the Huichang Era and never restored in later eras, which is discussed in Section 4. Only one treasure of Tang Jingzhongsi, the great bell, survived the persecution and later warfare and disasters until the Yongzheng Era of Qing, precisely because of the preciousness of its material. Yet, the stone-made antique images, most of which were fractured, show limited value for proper restoration and preservation. Based on the absence of records of any early images during the periods of destruction and reconstruction between the Song and mid-Qing periods, it can be inferred that the Buddhist hoard was “encapsulated” underground after the Huichang Persecution or possibly a little later, and remained separate from the changes occurring aboveground in the monastery, until the initial discovery of the hoard in 1882.

When the first “excavation” took place in 1882, the wording addressed in TRGZJ was “the former site of Wanfosi”. What happened to the monastery between 1714 and 1882? Zhang Zikai contends that the statement phrasing it as a “former site” (guzhi) in TRGZJ indicates the historical monastery was already completely destroyed sometime between 1714 and 1882 (Zhang 1999, p. 305). As aforementioned, the two pre-modern maps made, respectively, in 1879 and 1911 marked “Wanfosi” outside the old West Gate of the town (Figure 1). Therefore, I suggest that the “former site” of Wanfosi, as stated in TRGZJ, refers to some architectural structures that had remained in 1882, possibly including the main hall, which was rebuilt in 1741. Although Wanfosi was demolished sometime after its final restoration in 1714, the location of its exact site was identifiable in the late Qing in Chengdu. In the republic period, a map made in 1933 by the local department of the geodetic survey did not mark “Wanfosi” to the north of Jinhuaqiao (Figure 5) (Canmou Benbu Sichuan Ludi Celiangju 2012). As a result, it is reasonable to assume that the site was eventually completely destroyed sometime between 1882 and 1933 (Figure 5). Therefore, possibly, people involved in the 1937 excavation (or any undocumented digs after 1933) did not recognize the precise location of Wanfosi, but mistakenly identified the nearby Xiao Wanfosi as the former. This helps in understanding the different statement concerning the location of Wanfosi in SCNSZ, as discussed in Section 2.

4. The Hoard’s Formation

The formation of the hoard refers to how the stone images were assembled, preserved, and buried. Although there is a lack of any modern “scientific” archaeological context, the previous sections have displayed that all extant stone images from different excavations come from one single assemblage. Therefore, we are able to discuss this image assemblage as if it was a one-off excavation. As the Wanfosi hoard yields inscribed specimens that date to mid-Tang, one can quickly note that the final burial of the hoard occurred much later than, and seemingly irrelevant to, the pre-Tang eras. As a matter of fact, the pre-Tang images in the hoard and their burial in a much later phase were studied in isolation in previous research. Yet, I contend that the formation of the hoard bears essential clues concerning its “pre-Tang context”, that is, how the images were circulated, preserved, or even venerated during the pre-Tang periods. This is because the assemblage status could be, or could not be, consistent from the Liang to Tang periods. In other words, a later historical event assembling earlier Buddhist images is a possible and reasonable scenario. For example, Falin 法琳 (572–604) in his Bianzheng lun 辯正論 (Discerning the Correct) recorded a nationwide movement of repairing old Buddhist images (xiuzhi guxiang 修治故像) between 581 and 604, guided by the policy of renovating Buddhism of the Sui court.31 Therefore, we cannot naturally assume that the hoarded images cache was originally preserved in situ in the first place. The verification or falsification of this issue requires logical deduction, which was overlooked by the previous scholarships. As a result, it is necessary to first pay attention to the final burial of the hoard and, subsequently, regarding the chronological sequence of the dated specimens, I probe into their pre-Tang circumstances by specifically studying a Liang image re-dedicated in mid-Tang.

4.1. The Final Burial of the Hoard

The burial of the hoard pertains specifically to the most recently dated specimen within the entire collection. In north China, several other renowned Buddhist hoards were discovered after the 1950s, such as the Longxingsi 龍興寺 hoard in Qingzhou 青州 (Shandong), Xiudesi 修德寺 hoard in Quyang 曲陽 (Hebei), and Beiwuzhuang 北吳莊 hoard in Linzhang 臨漳 (Hebei). With scientific excavations already conducted, these hoards imply less ambiguous burial dates when compared with the Wanfosi hoard. For example, the Beiwuzhuang hoard formed no earlier than early Tang based on the inscriptions and a stratigraphy study, and it is probably associated with the Huichang 會昌 Persecution between 840 and 846 (Yecheng Kaogu Dui 2013, pp. 49–68). The Longxingsi hoard yields bronze coins and porcelains dating to the late Northern Song, and the burial date is framed no later than early Jin 金 (ca. mid-twelfth century) as per a stratigraphy study (Shandongsheng Qingzhoushi Bowuguan 1998, p. 14). In addition, the formation of the Xiudesi hoard was suspected to occur in situ during warfare in the An Lushan 安祿山 Rebellion (755–763), considering the latest yearmarks, literature records, and geographical characteristics of the excavation site (Luo 1955, pp. 37–38). Yet, our Wanfosi hoard is a site that has undergone multiple unscientific excavations, during which the original materials were partially or even largely looted or lost. However, it can be assured that there is no current evidence on post-Tang specimens of any kind reported from the site. A dhāraṇī pillar, excavated in 1951 from the site, bears a yearmark of 847, which is the latest recorded date among all of the inscriptions, and it should be the closest date to that of the burial date. This dhāraṇī pillar was carved with the Dhāraṇī of the Jubilant Buddha-Corona and the votive inscription with characters in the regular-running script (xingkai shu 行楷書) of extraordinary quality (Figure 6). The votive inscription is transcribed and translated as follows:

Huiguang, the person of great virtue of Zaixingsi. On the guimao day, the seventh day of the third month in the inaugural year of the Dazhong Era of the great Tang, the person of great virtue, the Headquarters Adjutant (Hucker 1985, p. 575, entry no. 7860) of the He Office of the Zhenjing Army,32 Probationary Chief Musician of the Court of Imperial Sacrifices (Hucker 1985, pp. 239, 431, 476, entry nos. 2477, 5204, 6145), Lord Yang Geng, spouse Lady Zhao, son Hongdu. Hongdu, at Jingzhongsi, built and erected one banner of the honored victor, for... of the dharma realm...

再興寺大德惠廣. 大唐大中元年叁33月柒日癸卯, 再興寺大德,34 鎮靜軍和衙官, 試太常寺協率郎楊公椩35, 妻趙氏, 男弘度. 弘度於淨眾寺建立尊勝幢壹所, 為法界...36

This dhāraṇī pillar was made only one year after the Huichang Persecution, during the Dazhong 大中 renovation period under Emperor Xuanzong’s 宣宗 reign (r. 846–859). The venue is clearly recorded in the inscription as “Jingzhongsi”, the in situ Tang predecessor of the Ming Wanfosi, the historical lineage of which was studied in the previous section (Figure 4). Jingzhongsi was a renowned monastery that was recorded in multiple sources of literature. The demolition during the persecution was recorded by Song Gaoseng zhuan (T50, no. 2061) as follows:

At first, in the abolishment of the teaching (of Buddhism), Chengdu preserved only one monastery, the Dacisi (Monastery of the Great Mercy). Jingzhongsi was discarded and destroyed without exception. The enormous bell of the monastery was thus moved into Dacisi. It was until the revival of the doctrine by Emperor Xuanzong, the bell was then returned to Jingzhong(si).

先是武宗廢教, 成都止留大慈一寺, 淨眾例從除毀. 其寺巨鐘乃移入大慈矣. 洎乎宣宗中興釋氏, 其鐘却還淨眾.37

Moreover, the renovation during the Dazhong Era can be referenced by Yizhou minghua lu 益州名畫錄 (A Critique on Famous Paintings of Yizhou) as follows:

Fan Qiong... Along with Chen Hao and Peng Jian, who lived in the same era and (had) the same expertise, temporarily lodged in the city of Shu... During the Huichang years, after the destruction, only one monastery, the Da Shengcisi, preserved the Buddhist images. It was until the renovation of Buddhist monasteries by Emperor Xuanzong, the three individuals, at Shengshousi, Shengxingsi, Jingzhongsi, Zhongxingsi, from the Dazhong Era to the Qianfu Era, (having their) brush-pen never temporarily eased, pictured and painted more than two hundred jian of walls.

范瓊者... 與陳皓, 彭堅同時同藝, 寓居蜀城... 會昌年除毀後餘大聖慈一寺佛像得存. 洎宣宗皇帝再興佛寺, 三人於聖壽寺, 聖興寺, 淨眾寺, 中興寺, 自大中至乾符, 筆無暫釋, 圖畫二百餘間墻壁...38

As can be seen, in the Huichang Persecution, Jingzhongsi did not escape the destruction during the campaign. Yizhou minghua lu mentions only that the Da Shengcisi (aka. Dacisi 大慈寺) preserved its Buddhist images, indicating that other monasteries in Chengdu, including Jingzhongsi, failed to preserve theirs. It can be deduced that there were originally mural paintings in Jingzhongsi because the above passage reports that three painters had repainted areas during the renovation. The demolition of mural paintings involves such brutality as burning and chiseling. Naturally, the free-standing images in the monastery also underwent devastating atrocities during the campaign. Therefore, the wording addressed in the above passage, the “Buddhist images” (foxiang), should also include free-standing images. However, excluding the dhāraṇī pillar dated to 847, all other dated materials from the hoard are from the pre-Huichang periods. One reasonable assumption is that the primary components of the hoard did not survive the persecution campaign between 840 and 846. This helps to explain the Buddha heads’ unexceptional separation from their torsos and the fragmented condition of some of the Tang pieces. Following this, the dhāraṇī pillar should have been buried after the persecution campaign, supposedly even after a long run during the revival of the monastery. In this case, the Dazhong campaign did not restore the previously disposed of free-standing images. Another scenario is, in contrast, the hoard materials survived the persecution and were buried along with the dhāraṇī pillar in a later period. As one quickly notes the masterly carved dhāraṇī pillar and its high-graded devotees, as well as the Painter Fan Qiong’s restoration, Jingzhongsi should have recovered its prosperity swiftly after the persecution. Did the Dazhong campaign also involve the restoration of the disposed free-standing images? By closely examining the Tang specimens for any traces of repairs, we might gain insights into potential treatment after the Huichang Persecution. However, the available images and information at present are insufficient to provide valuable clues regarding this matter. Nonetheless, considering it as a plausible hypothesis, we should continue to establish the terminus ante quem of the final burial.

After the Huichang Persecution, several upheavals took place in the late Tang. The first was the Huang Chao’s 黃巢 Rebellion between 875 and 884, during which Emperor Xizong 僖宗 of Tang (862–888, r. 873–888) relocated to Chengdu in 881. During this period, the Shu 蜀 region constantly experienced unrest, such as the Qian Neng 阡能 (d. 882) Rebellion between 882 and 883. Shortly after Emperor Xizong’s death, between 888 and 891, the rebellious Chen Jingxuan 陳敬瑄 (d. 893) struggled against the Tang force in Chengdu for three years, which led to a food shortage, child abandonment, deaths from starvation, and other atrocious cruelties.39 Subsequently, the Former Shu前蜀 (907–925) regime established by Wang Jian 王建 (847–918) was engaged in multiple wars against surrounding polities, especially the warlord Li Maozhen 李茂貞 (856–924) in the north. In 924, the Former Shu was defeated by Later Tang 後唐 (923–937), established by Li Cunxu 李存勖 (885–926), during which the last ruler, Wang Yan 王衍 (899–926), capitulated in Chengdu. Afterward, Chengdu experienced chaos during the takeover by Later Shu 後蜀 and war with the northern regime Later Zhou 後周 (951–960). The chaos of warfare did not cease even after the takeover of Chengdu by the Northern Song 北宋 (960–1127) power in 965, and a durable anti-Song confrontation centered in Chengdu lasted approximately for thirty years until ca. 1000, which caused economic collapse, robbery, and even slaughter.40 Overall, the burial of the hoard materials could have occurred during the continuous chaos between the Huichang Persecution and the early takeover of Northern Song. Afterward, Chengdu experienced a durable prosperous period, reflected in its population growth, economic development, and good administration. However, not a single inscribed stone, stoneware fragment, nor bronze coin of the Northern Song was reported from the site. Therefore, it does not seem reasonable to suspect a terminus ante quem later than the end of the first millennium.

Above all, we have framed the burial date of the hoard as between 840 and 1000. Two possible scenarios for the final burial can be suggested. First, if the main part of the hoard materials were buried during the Huichang Persecution, the Tang dhāraṇī pillar would have been disposed of during a second burial in the coming centuries. Otherwise, the hoard was formed in a one-off burial sometime in post-Huichang periods and no later than ca. 1000. In either case, a crucial insight concerning the earlier context remains unchanged, that is, the earlier stone images still existed as an assemblage until late Tang. The status of the gathered images contributes to our understanding of the hoard’s earlier circumstances.

4.2. Brief Chronology of the Hoard

There are 12 dated specimens in total from the Wanfosi hoard as shown in Table 1. Among these images, six bear Liang yearmarks: two life-sized standing Buddhas respectively dated to 529 and 537 (Table 1, nos. 3, and 5, Sichuan Bowuyuan et al. 2013, pp. 20–21, 32–33), and four group-configured images respectively dated to 523, 525, 533, and 548 (Table 1, nos. 1, 2, 4, and 6) (ibid., pp. 77–81, 82–85, 86–90, 91–94). Two images are inscribed with Northern Zhou yearmarks: a life-sized standing Buddha with an era mark dated between 562 and 565,41 and a life-sized seated bodhisattva dated to 567 (ibid., 74–76). Following the Northern Zhou period, although there are no Sui 隋 (581–619) specimens currently extant, the earliest antiquarian record from Wang’s TRGZJ had documented a now-lost stone bearing an eramark of the Kaihuang 開皇 (581–600) of Sui (Table 1, no. 13).42 It is reported that the Sichuan Museum collected 22 specimens from the Tang period in the hoard (He and Dong 2014, pp. 63–66).43 Among these materials, one “stone image” bears a yearmark of 728 (Table 1, no. 9),44 a “bodhisattva image” bears a yearmark of 737 (Table 1, no. 10),45 and the abovementioned dhāraṇī pillar bears a yearmark of 847 (Table 1, no. 12).46 Unfortunately, for the two Tang specimens dated to 728 and 737, no photographs have yet been published, to the best of my knowledge. The descriptions in CWSZ and CWSY are too inadequate to be matched with the museum’s collection via accession numbers. A special image bears a Tang inscription from 815 (Table 1, no. 11, Sichuan Bowuyuan et al. 2013, pp. 95–98), but it cannot be a Tang product according to a stylistic analysis, which is discussed critically in the next subsection.

To more intuitively perceive the chronological order of the dated images from Wanfosi, all 13 specimens (including one textual record from TRGZJ) are illustrated in Figure 7, which references the two Buddhism persecutions in Northern Zhou and Tang, and the estimated final burial date. My initial impression is that, although there is an enduring span of 324 years between the earliest dated specimen and the latest one, the materials from the hoard are distributed in several successive historical periods from Liang to Tang. This phenomenon helps in understanding the “uninterrupted narrative” between the Tang monastery and its pre-Tang complex, which is discussed critically in the following subsection.

4.3. A Re-Used Image Living through Liang and Tang

This subsection places special emphasis on an intriguing reused stone image that plays a significant role in enhancing our understanding of the materials comprising the Wanfosi hoard’s formation. The image in question features an inscription from the tenth year of the Yuanhe 元和 Era of the Tang Dynasty (815), without preserving the name of the devotee, and therefore I address it as the “Yuanhe Image” (Table 1, no. 11). On the reverse side, the votive inscription is in a deteriorated state. It has been transcribed and translated as follows (Figure 8):

On the sixth day of the eighth month, in a yiwei year, the tenth year of the Yuanhe Era of the Great Tang (815)... Assembly of... on purpose that... deceased... deceased... save... The entire family, the old and young, Qingji respectfully made one body of image of Buddha Śākyamuni, making offerings in front of [Jingzhongsi]. May the deceased ascend to the heavenly realm(s), and the entire family enjoy such a blessing... (Buddhist) Society Member... is one... in the heavenly realm... wife... Zhou.

大唐元和[拾]年/太歲在乙未八/月朔六日[ ]眾/[ ]等為[亡][ ][ ]/[ ][亡]保合[家]大小/[清]吉敬造[釋][ ]佛/像一[區]供[淨][眾][寺]/前願[亡]者昇天合/家咸[享]斯福/[ ]邑子[ ]乃[一]....../......[在]天[媳][ ]/......周.

(ibid., p. 98)

The votive inscription of the Yuanhe Image seemingly bears the name of a location in the seventh row after the character gong (“making offerings”). These three characters were transcribed for the first time in SCNSZ with an “uncertainty mark”(ibid.). Upon careful recognition of the carved strokes (Figure 9), although largely damaged, one can identify the double-drips (or triple-drips) radical on the left and the upper-right structure that looks like a ji 彐 of the first character, and the upper structure that assembles the character si 四 of the second character. The third character is si 寺 without much uncertainty. These traces do agree with the charters of “Jingzhongsi”, Tang monastery complex in situ, as already discussed in previous sections.

Regardless of its inscription, as briefed beforehand, the Yuanhe Image, in any case, cannot be a product of Tang in terms of its artistic style. There are six images in a similar group-configuration style in the Wanfosi hoard, four of which are rather complex, showing a total of up to 11 figures (Figure 10, nos. 1–4). All four images bear Liang yearmarks, as is also demonstrated in Table 1. The configurations are consistent: a main figure (mostly Śākyamuni, although one case is Avalokiteśvara according to the inscription), four flanking bodhisattvas on two sides of the main figure, four disciples carved in low relief on the aureole background, and two guardians on both sides on the front. These images also share other similar formal characteristics. Their aureole backgrounds are forwardly bent and in the shape of the outline of a frontview peach (or a large lotus petal). The carving on the aureole background is divided into inner and outer layers (also the innermost layer if the halos of the figures are counted as well). Without exception, the layers are divided by stringed beads. In addition, the ranking and position of different identities and body postures of the flanking figures, and the two energetic playful lions in front of the throne, are broadly consistent. Beyond the Wanfosi hoard, these characteristics also exist in other Liang group-configured images in other hoards excavated in present-day urban Chengdu. The Yuanhe Image’s formal characteristics (Figure 8), including the forward-bent aureole background, the figure composition, and the three-layered diagram on the background of the obverse, together with each figure’s artistic execution, have been eloquently stated its Liang attribute despite its Tang inscription. The original carvings and inscriptions on the reverse of the Yuanhe Image were likely effaced and polished for a later inscription in the Tang period.

The Yuanhe Image exhibits several indications of reuse. One notable indication is the relatively thinner aureole background when observed from the side, in comparison to other group-configured images found in Chengdu. Usually, there used to be a devotional scene in the upper part and a tablet for the inscription below, which were polished away for the new inscription. There could have been a guardian on each side view, which were probably removed during the renovation because of the destruction to the stele’s sides when polishing the entire back of the stele. In addition, there is a trapezoid-shaped sunken surface on the top edge on the reverse side (Figure 9, upper right). This should have been functioning for inlaying a tenon (shape assembling the “dovetail tenon” in wooden craft practice) to repair the broken upper part of the aureole background. These tortuous treatments of an antique stone image show a very dedicated devotional religious practice from a much later era. Additionally, the engraved caption ”Buddha Śākyamuni“ (Shijiamouni wenfo 釋迦牟尼文佛) on the reverse side is missing the first character “shi” due to the absence of the upper-top part. Due to the simultaneous engraving of the caption and the dedicatory inscription during the process of refacing and rededicating, it is expected that the Yuanhe Image, which is currently missing the upper-top part, would have remained intact and well preserved in 815.

Hence, we can learn one critical fact from this re-use practice. In the early ninth century Jingzhongsi of Tang, there was (or probably were), admittedly, Liang Buddhist image(s) dated to some three centuries ago, which had still been practically used for devotional purposes. The production of new images in Jingzhongsi was initiated by Monk Musang, and the two images dating to 728 and 737 (Table 1, nos. 1, 9, and 10) are evident of the image-making campaign in the early stage. The Yuanhe Image did not seem to have circulated among different households or monasteries in pre-Tang eras. This conclusion is drawn from the fact that the image managed to survive the Jiande 建德 Persecution (574–578), and all other challenging situations such as warfare and natural disasters, throughout the course of three centuries. Remarkably, it remained in relatively good condition until 815. Moreover, it is worth noting that the Yuanhe Image existed alongside other Liang images within the hoard. It would be almost unreasonable to assume that an early ninth-century monastery had acquired scattered antique images dating back three centuries ago from various locations in Chengdu. Therefore, I suggest that the Yuanhe Image was perpetually kept at one venue, and it had never circulated outside the monastery. This perpetual shelter for the preservation of stone images was the long-lasting monastery Jingzhongsi of Tang and its pre-Tang complex. We have limited knowledge about the pre-Tang history of Jingzhongsi, but the image cache was preserved until Monk Musang established the Tang monastery. Over time, newly crafted images were added to the existing collection, ultimately forming what is now known as the “Wanfosi hoard”. After all, this rare example of a reused Liang image in Tang has allowed for the logical deduction of its Pre-Tang situation because of the “uninterrupted narrative” throughout the entire period (green dashed line in Figure 7). Although the final burial happened between late Tang and early Northern Song (ca. 840–1000), the assemblage status of the hoard, or what we currently perceive as the “Wanfosi hoard”, had been already formed, long before the final burial.

5. Conclusions

This article discussed a collection of Buddhist images that, unlike a modern archaeological excavation, accumulated through time due to at least four major intermittent excavations. The available written information concerning these excavations is very limited. The excavation site linked to the ruined monastery Wanfosi has sparked inquiries regarding its exact location. Of particular interest is the contradiction found in the most comprehensive hoard catalog SCNSZ, which provides a different account of Wanfosi’s present-day whereabouts, which is inconsistent with the pre-modern sources. This deviation could have led to a basic question: whether it was one integrate hoard or two different hoards instead. In Section 2, upon careful review of the extensive sources of the four unsystematic excavations, I proposed that the dispute of the location might have been caused by a relocation event that is recorded in TRGZJ. Therefore, the excavated materials from four separate excavation campaigns, namely, those conducted in 1882, 1937, 1945/1946, and 1953/1954, should originate from a single location. This location is presently identified as the present-day hostel area of the present-day China Railway No. 2 Group in downtown Chengdu. While my conclusion aligns with the commonly held belief that the images originated “from the former site of Wanfosi”, my observation of two distinct locations and their connection through an “image relocation” event should not be disregarded. This step is crucial as it not only provides a precise location for the hoard but, more importantly, confirms that the multiple batches of images obtained through “unscientific” excavations indeed belong to a single hoard.

In Section 3, this article explored the historical lineage of the monastery closely associated with the Wanfosi hoard, with a particular focus on gazetteer records. The namesake monastery “Wanfosi” was one of the previous names of the historical Jingzhongsi of Tang. The production of images in Jingzhongsi experienced a flourishing period starting in the Musang Era and continued to thrive until the Huichang Persecution. While the monastery underwent general restoration during the Dazhong Era, certain sections remained unrestored. It was likely during the Huichang Persecution or a slightly later period that the Buddhist hoard was buried underground, detached from the changes occurring within the aboveground monastery. It remained dormant until its initial discovery in 1882. Subsequently, the site underwent undocumented deterioration and was eventually left in ruins sometime between 1882 and 1933. Consequently, it is possible that individuals involved in the 1937 excavation (or any undocumented excavations after 1933) failed to identify the precise location of Wanfosi. Instead, they mistakenly associated the nearby Xiao Wanfosi as the former site.

Section 4 began with a discussion of the hoard’s final burial date, primarily examining the latest artifact, a dhāraṇī pillar, and considering potential events that could have led to its burial. Based on this analysis, the burial of the hoard is estimated to have taken place between 840 and approximately 1000. Subsequently, after summarizing a chronological sequence of the inscribed dates of the hoard, a particular focus was given to the Yuanhe Image, which has received minimal attention in previous research. This unique image sheds light on the scenario of the Tang Jingzhongsi preservation of antique Liang Buddhist images for devotional purposes. I suggest that the preservation of this image, along with other excavated Liang images, likely occurred long before the final burial of the hoard. This said, the assemblage of the hoard had already taken shape well before its ultimate burial. Although we have limited knowledge of the pre-Tang history of Jingzhongsi, the image cache was preserved until Monk Musang established the Tang monastery. Over time, newly made images were added to the existing collection, ultimately forming what is now known as the “Wanfosi hoard”.