Praṇidhi Paintings and Inscriptions of Cave 20 at Bezeklik and the Mūlasarvāstivāda-Vinaya Bhaiṣajyavastu

Abstract

The praṇidhi paintings of Cave 20 at Bezeklik in Turfan are murals representing the cultural sophistication of the Gaochang Uyghur kingdom (866–1283). Building on Grünwedel’s rearrangement of the murals in his 1924 book, this paper examines praṇidhi paintings of Cave 20, with a particular focus on analyzing the accompanying inscriptions in relation to Mūlasarvāstivāda-vinaya Bhaiṣajyavastu. Le Coq’s numbering system, used in existing literature, has made it difficult to comprehend not only the sequential arrangement among 15 themes but also the order in which inscriptions were derived from the text. Using an alternative arrangement of the paintings, this paper provides a systematic pattern in the derivation of inscriptions from the text, with the following results: The inscriptions of Subject 9 (Ratnaśikin Buddha), Subject 7 (Dīpaṃkara Buddha), and Subject 10 (Kāśyapa Buddha) correspond respectively to three asaṅkhya kalpas of the verses, each conveying a pivotal moment in its period. Seen from left toward the right corridors, the inscriptions follow a sequence tracing from the second to the third, and then the first asaṅkhya kalpa. Following the pictorial narrative, the cycle begins with inviting the Buddha, the donor’s offering and worship, continues with making vows and receiving prophecies from the three Buddhas, and concludes with Buddha being seen off and renewal of devotional engagement by the viewer. These findings help clarify an issue in previous scholarship concerning how the inscriptions correspond to the Mūlasarvāstivāda-vinaya Bhaiṣajyavastu. This also supports Grünwedel’s early insight that the praṇidhi paintings of Cave 20 were interconnected with circumambulatory religious practice.

1. Introduction

The praṇidhi paintings of Cave 20 at Bezeklik in Turfan are murals representing the cultural sophistication of Gaochang Uyghur kingdom (866–1283). The murals hold significance in understanding the history, religion and religious art of this period (Meng 1981). This paper examines praṇidhi paintings of Cave 20, with a particular focus on analyzing the accompanying inscriptions in relation to Mūlasarvāstivāda-vinaya Bhaiṣajyavastu—the chapter on medicine in the monastic code of the Mūlasarvāstivāda school. Noting the difficulty of existing literature in comprehending the sequential arrangement of the paintings and the order in which inscriptions were derived from the text, I argue that these challenges can be attributed to the literature’s reliance on Le Coq’s numbering system (Le Coq 1913). As an alternative, using Grünwedel’s rearrangement of the murals in his 1924 book, this paper identifies a systematic pattern wherein the inscriptions are derived from Mūlasarvāstivāda-vinaya Bhaiṣajyavastu.

The praṇidhi paintings in Cave 20 are large-scale murals exceeding 3 m in height, consist of 15 subjects, and are located on both the inner and outer walls of the cave corridor. They all depict a large standing Buddha at the center of their rectangular frames, with donors of various social statuses, such as kings, Brahmins, and merchants, positioned at the lower left and right, surrounded by deities, monks, and Vajrapani. The murals are visually unified by drapery above and brick motifs below, while individual scenes are thematically delineated by vertical decorative borders. The scenes feature a red background, with natural landscapes depicted at the bottom and inscriptions at the top. These murals are generally interpreted as depicting scenes in which Śākyamuni, in his previous lives, makes offerings and vows to the past Buddhas and receives prophecies from them.

However, despite their iconographic coherence and integrity, the murals themselves have not survived to the present day. The praṇidhi paintings of Cave 20 were cut from the cave wall and removed by Albert von Le Coq (1860–1930) during his second German expedition in 1905 and subsequently exhibited at the Museum of Ethnography in Berlin. The paintings were destroyed during World War II and have since been lost. The plates published in Le Coq’s book in 1913, which document both the paintings and their accompanying inscriptions, remain the primary source for the study of the praṇidhi paintings. As the book remains the only visual source for research on Cave 20, and the murals were severed from their original context, there is an inherent limitation in fully understanding the narrative of the murals.

Notwithstanding the loss of the murals, the praṇidhi paintings and their inscriptions of Cave 20 have drawn significant scholarly attention, particularly with regard to their possible connection to the Mūlasarvāstivāda. The inscriptions, written within a cartouche at the upper part of the scene in Sanskrit using Brāhmī script,1 have long been the subject of scholarly study. In 1913, Le Coq, with the assistance of Wilhelm Siegling, published German interpretations of the 11 subjects in Chotscho (Le Coq 1913, pp. 17–29). In the same year, Heinrich Lüders interprets the inscriptions of 14 subjects in German, excluding Subject 12, whose inscription had already been damaged at the time of Le Coq’s expedition (Le Coq 1913, p. 15). Lüders notes the expression of the three asaṅkhya kalpas2 (incalculably long eon) in the inscriptions and points out that the same expressions also appear in the Dharmarucyavadāna of Divyāvadāna, which confirms that the Sarvāstivāda3 distinguishes three asaṅkhya kalpas (Lüders 1913, pp. 261–63). Taking all evidence together, he suggests that the inscriptions and praṇidhi paintings of Cave 20 may belong to the Sarvāstivāda lineage. Lüders further assumes that if texts of praṇidhicaryās (vow practices) from the Sarvāstivāda were found in Sanskrit manuscripts or their Chinese and Tibetan translations, they would align much more closely with the inscriptions on the murals. In the following year, Edouard Huber found a Chinese translation of the Mūlasarvāstivāda-vinaya Bhaiṣajyavastu (hereafter Bhaiṣajyavastu), in which some verses correspond to the inscriptions of praṇidhi painting subjects 2, 3, 5, 6, and 10 (Huber 1914, p. 12).

In a subsequent study, Shinkan Hirano4, in his 1961 paper, interprets the inscriptions in Japanese following Le Coq’s numbering sequence of the praṇidhi paintings and identifies passages in the Chinese and Tibetan translations of the Bhaiṣajyavastu corresponding to all subjects except subject 4 (Hirano 1961). He further details the inscriptions of the praṇidhi paintings and puts forward an argument that they reflect the Buddha-view and concept of praṇidhi5 in the Vinaya of the Mūlasarvāstivāda (Murakami 1984). It should be noted that Murakami examines the ordering in which the inscriptions of the praṇidhi paintings appear in the verses of the Bhaiṣajyavastu, mentioning that he could not find any systematic order showing the relationship between the inscriptions and the Bhaiṣajyavastu (Hirano 1961, p. 123; Murakami 1984, p. 254).

Recently, Tianshu Zhu and Haewon Kim also explored the relationship between the inscriptions and the verses of the Bhaiṣajyavastu, reaching a similar conclusion to Murakami that the order of extraction of inscriptions from the verses is not identifiable (Zhu 2012, pp. 76–79; Kim 2013, pp. 36–37). Given that the inscriptions of the praṇidhi paintings originate from the Bhaiṣajyavastu and reflect the thought of the Mūlasarvāstivāda, one may wonder why scholars have found it difficult to identify a pattern demonstrating how the inscriptions were selected from the verses. In this respect, this paper suggests considering the possible effect of Le Coq ‘s numbers on existing literature. Specifically, Le Coq’s numbering begins from the left corridor, first assigning numbers to the murals on the outer wall, and then gives the following numbers to murals on the inner wall (Le Coq 1913).6 I believe that this numerical order for the praṇidhi paintings does not adequately consider the three-dimensional space of the corridor-style cave where the murals were situated, nor the natural sequence in which viewers were likely to see them. As a result, Le Coq’s numbering system has made it difficult for researchers to comprehend not only the sequential arrangement among 15 murals but also the orders that inscriptions were derived from Bhaiṣajyavastu, which has in turn led recent research to focus on individual 15 praṇidhi scenes separately without considering a coherent narrative sequence for the scenes.

Cave 20 of Bezeklik is regarded as a royal monastery of the Gaochang Uyghur Kingdom, which flourished between 866 and 1283 (Meng 2020, p. 593). Given that the praṇidhi paintings are murals exemplifying the cultural sophistication of the Gaochang Uyghurs, they were likely planned and carried out in a highly intentional manner. Considering this, the 15 scenes of praṇidhi paintings likely form an integrated visual program within the architectural structure of the cave corridor (Figure 1).

Accordingly, I rearranged the praṇidhi paintings and investigated their inscriptions in relation to the verses of the Bhaiṣajyavastu. As an alternative approach, the arrangement of the murals proposed here builds on the ordering system suggested by Albert Grünwedel, with a few modifications to his original sequence (Grünwedel 1924). The structure of this paper is as follows: Section 2 examines the iconography and inscription content of the praṇidhi paintings according to Le Coq’s numbering sequence and highlights the problems in the arrangement of the praṇidhi paintings. Section 3 presents a rearrangement of the murals and analyzes the relationship between the inscriptions of the praṇidhi paintings and the verses of the Bhaiṣajyavastu.

2. Arrangement of Praṇidhi Paintings According to Le Coq’s Number

Figure 2 illustrates a floor plan of Cave 20,7 which is plotted based on the author’s fieldwork in 2007 and Le Coq’s floor plan in Chotscho (Le Coq 1913, p. 14). The numbers from ① through ⑮ represent the subject numbers of the praṇidhi paintings presented by Le Coq. As illustrated in this floor plan, following the sequence of numbers, viewers are assumed to first see the murals on the outer wall of the left corridor (1 → 2 → 3) and then return to view paintings on the inner wall of the same corridor (4 → 5). This pattern is repeated in the rear and right corridors. However, such a viewing path is not consistent with an arrangement of murals that follows the pradakṣiṇa—that is, circumambulating a stupa in a clockwise direction. To highlight the issues in interpreting the relationships among iconography, inscriptions and the Bhaiṣajyavastu,

Table 1. Arrangement of praṇidhi paintings according to Le Coq’s numbers: iconography, inscription and Bhaiṣajyavastu.

Subject Nos. in Cave 20	Iconography	Inscriptions	Verses in the Tibetan Bhaiṣajyavastu
①	The brahmin offers food (or incense) and invites the Buddha.	The worldly leader, Mahendra, was served by me, a brahmin, with a steam bath, various incense, and black agaru. Thereafter, having built a monastery, I invited the Buddha with all reverence.	I served the leader of the world, Narendra, by building a steam bath, and with the perfume of black agaru, When I was a brahmin.
②	The king offers an umbrella and prays to the Buddha.	Tamonuda, the greatly fortunate one, was venerated by me, a king, and was honored with a variety of jewels and musical tributes.	When I was a king, I offered veneration in the form of various treasures and excellent music to the fortunate one Tamonuda.
③	The king shaves his head and prays to the Buddha.	In a very pleasant city, Śikhin, the famous enlightened one, was venerated by me, a merchant, with many monasteries.	In the very city Delight, When I was the head of a guild, I venerated by building a monastery the Buddha Śikhin, who was renowned in the world.
④	The king offers food (or incense) and invites the Buddha.	Kṣemamkara, the Sun of humans, was venerated by me, who was a king, with a steam bath, various incense, and black agaru. Moreover, he was invited with sixty thousand monasteries.	When I was a king, in sixty thousand cities I invited and venerated the sun among humans, Kṣemaṃkara.
⑤	The king offers an umbrella and prays to the Buddha.	Ananda! Siṃha, possessing the strength of a lion and a hero among humans, was venerated by me, the king of men, with a jewel-handled umbrella.	Ānanda, as the master of the people I venerated Siṃha, who had a lion’ s power, that most excellent of humans, with an umbrella with a jeweled shaft.
⑥	The merchants offer their goods to the Buddha.	To venerate the six victors with elephants, horses, gold, women, treasures, and pearls, a grove was built by a merchant.	When I was the head of a guild, I built a park adorned with elephants, horses, gold, women, and jewels to venerate six victors.
⑦	The young brahmin offers lotus flowers and lets his hair down to allow the Buddha to pass.	Having seen the Dīpaṃkara radiant and renowned, the young brahmin venerated him with seven blue lotus flowers. At the end of the second asaṅkhya [kalpa].	When I was a young brahmin, I saw the Buddha Dīpaṃkara, who was famed for his radiance, and venerated him with seven blue utpalas.
⑧	The king offers a Cintāmaṇi to the Buddha.	Sunetra, the worldly leader, was at that time venerated by me, a merchant, with Cintāmaṇi jewels and a beautiful monastery.	When I was the head of a guild, I venerated the leader of the world, Sunetra, with precious jewels and a pleasant monastery.
⑨	The monk and the king offer a lamp to the Buddha.	Formerly, in another life, I was a princess. The oil of a lamp was offered to my brother Ratnaśikin. At the end of the first asaṅkhya [kalpa].	In another past life when I was a son of a king, I venerated my brother, Ratnaśikhin, with a donation of lamps.
⑩	The young brahmin prays to the Buddha while receiving the prophecy.	I was Uttara, a young brahmin. Having heard the words of Nandipāla, I resolved to become a monk under the Kāśyapa, the perfectly enlightened one. At the end of all virtuous practices of the third asaṅkhya [kalpa].	When I was the young brahmin Uttara, having heard Nandīpāla’s words, I formed a resolution to go forth in front of Kāśyapa, the best of humans.
⑪	The monk offers a robe to the Buddha.	Upon hearing of the arrival of Vāsiṣṭha, I, the merchant, was filled with joy. I decorated the grove and had a monastery built.	When I was the head of a guild, pleased when I heard that Vāsiṣṭha was coming, I adorned a park and built a monastery.
⑫	The king shaves his head. *	For the sake of Kanakamuni, Shakamuni, who was a merchant, built a grove. A monk……for the sake of great enlightenment……. **	When I was the leader of a guild, for the completely awakened one Kanakamuni I built a monastery and went forth into homelessness with faith.
⑬	The brahmin offers leopard-print cloth to the Buddha. *	I, who was indeed a ṛṣi, served Sunetra, the master of the three worlds. The buddha was dressed in a pleasant bark garment by me.	I honored Sunetra, the leader of the world, when I was a ṛṣi. I dressed that most excellent of humans with comfortable bark
⑭	The merchants pray and make offerings to the Buddha as he crosses the river by boat.	I saw Aṅgirasa arrive at the riverbank, and I, who was the merchant leader, ferried the buddha across the river.	When I was a ferryman, I saw Aṅgaratha arrive at the bank and helped the Muni Cross the water’s current.
⑮	The brahmin offers food (or incense) to the Buddha and sees him off.	Lawfully, Pūrṇa Manorata, was worshipped with faith by me, who had become king, and I was then honored with the title of Dharma King.	When I was a king, I venerated correctly, out of faith, the Buddha Pūrṇamanoratha, and I was called a Dharma king.

Sources: Author’s own compilation; Le Coq (1913); Andrews (1948); Murakami (1984); 84000 Translation Project (2021). Notes: The numbers in column 1 from ① through ⑮ refer to the subjects of the praṇidhi paintings numbered by Le Coq. * Indicates the subjects that no longer remain. Since Subjects 12 and 13 of Cave 20 have not survived, they are explained by referring to the corresponding subjects in the same position in the corridor of Cave 15, as well as Le Coq’s description in Chotscho (Le Coq 1913, p. 15). ** Denotes the inscriptions that no longer remain. I insert a fragment of inscription, which was discovered at Sengin Temple No. 1 and suggested by Murakami’s book in 1984, as a corresponding piece to Subject 12 (Murakami 1984, pp. 236–37).

summarizes their content according to Le Coq’s sequence. The details of inscriptions are drawn from the Japanese translation interpreted by Murakami’s book, while the Tibetan Bhaiṣajyavastu verses are taken from its English translation (Murakami 1984; 84000 Translation Project 2021).

Assuming that a viewer would proceed along the corridor according to Le Coq’s sequence, the pictorial program should begin from the outer wall, which are ① ② ③ according to Figure 1a, at the entrance of the left corridor (see Figure 2). As shown in column 2 of Table 1, Subject 1 depicts a brahmin making offerings to the Buddha and inviting him. On the adjacent outer wall, Subject 2 shows a king making umbrella offerings and praying to the Buddha, and Subject 3 depicts a king having his head shaved and praying to the Buddha. Following this sequence, the viewer must then retrace their steps back toward the corridor entrance in order to view Subject 4 on the inner wall. However, its scene is a king making an offering and inviting the Buddha, a theme similar to that in Subject 1. The scene of Subject 5 again shows a king making umbrella offerings and praying to the Buddha, as seen in Subject 2 (see Table 1 and Appendix B). One can easily notice that such a viewing path is unnatural, as it requires the viewer to move back and forth within a corridor barely 1.2 m wide.

In addition, if this viewing path is extended into the rear corridor, the viewer would encounter Subjects 6, 7 and 8 on the outer wall of the rear corridor, followed by Subjects 9 and 10 on the inner wall. As shown in column 3 of Table 1, the viewer first encounters the subject of offering lotus flowers to Dīpaṃkara, which is Subject 7 from the second asaṅkhya [kalpa]. This is followed sequentially by the subject of offering lamps to Ratnaśikin, Subject 9 from the first asaṅkhya [kalpa], and the subject of receiving the prophecy under Kāśyapa, Subject 10 from the third asaṅkhya [kalpa]. As a result, the order (7→9→10) produces another puzzling sequence of second, first, and third asaṅkhya kalpa.

Furthermore, Le Coq’s sequence leads us to a mystery regarding the order of the inscriptions that are derived from the Bhaiṣajyavastu. For this reason, there is an argument that “in the praṇidhāna paintings of such stories the sequence and the selection of the themes appear to be somewhat disordered and even chaotic” (Zhu 2012, pp. 78–79).

However, it should be noted that Subjects 1 (Figure 3a) and 4 (Figure 3b) have considerably similar inscriptions and iconographic content and are positioned facing each other at the entrance of the left corridor of the cave, which allows us to consider their paired relationship. On the left corridor, Subject 1 is on the outer wall, and Subject 4 is on the inner wall, positioned on the facing wall. Both subjects commonly depict a brahmin and a king inviting the Buddha; both kneel on their knees, gaze at the Buddha, and extend their hands forward with a gesture of invitation. In both scenes, the figures standing behind the Buddha—the brahmin and brahmin boy in Subject 1, and the king and queen in Subject 4—commonly hold trays of offerings, thereby forming a pair in that corridor. The inscriptions of these two subjects are also similar (column 3 of Table 1), indicating the brahmin and king who offer bathing and incense, and invite the Buddha to the monastery.8 All of these clearly suggest a discernible visual scheme connecting Subjects 1 and 4.

Similarly, one can further notice iconographical and contextual similarities between Subjects 2 (Figure 4a) and 5 (Figure 4b). Subjects 2 and 5 are situated immediately adjacent to Subjects 1 and 4 and likewise arranged to face each other.9 These two scenes depict a king offering an umbrella and praying to the Buddha. In both scenes, positioned in the direction of the Buddha’s gaze, the king kneels with his hands joined in añjali gesture while worshipping the Buddha. In Subject 2, a king is shown at the lower left holding up an umbrella as an offering, while in Subject 5, a king and queen stand side by side at the lower right with an umbrella in hand, illustrating the same iconographic theme.

In sum, Subjects 1 and 4 begin with presenting offerings and inviting the Buddha, and in the following scenes, Subjects 2 and 5, the donors hold umbrellas and worship the Buddha, thus sequentially conveying the common iconographic theme. Viewed in this light, Le Coq’s sequence obscures the visual scheme and coherent narrative structure of the subjects, thereby leading to doubt on the validity of his numbering system.

3. New Arrangement of Praṇidhi Paintings and Their Relationship to the Bhaiṣajyavastu

Grünwedel, in his study of the praṇidhi paintings of Cave 20, notes that the movement of the pradakṣiṇa is the central concept (Grünwedel 1924, pp. II 79–92). He accordingly interprets the sequence of the praṇidhi paintings in the following order: Subjects 1, 4, 2, 5, 3 → Subjects 6, 9, 7, 8, 10 → Subjects 11, (12), 14, 13, 15 (see Figure 2). Building on this approach, I rearrange the paintings by considering both the path of the corridor, where the viewer would see the outer and inner walls simultaneously, and the order in which the inscriptions of the murals appear in the verses of the Bhaiṣajyavastu. The sequence is as follows: left corridor Subjects 1, 4, 2, 5, 3 → rear corridor Subjects 6, 9, 7, 10, 8 → right corridor Subjects 11, 14, 12, 15, 13.

The table below,

Table 2. Comparison of the relationships among praṇidhi painting arrangement, iconography, inscriptions, and the Bhaiṣajyavastu.

Placing of Subject in Cave		Sequence of Subject	Iconography	Inscriptions			Relationships of Iconography and Inscriptions	Order of Inscriptions in the Bhaiṣajyavastu
				Former Status of the Donor	Past Buddha	Methods of Offerings
Left corridor	Outer	①	The brahmin offers food (or incense) and invites the Buddha.	Brahmin	Mahendra	Steam bath, Incense, Monastery	Consistent	7th—II
	Inner	④	The king offers food (or incense) and invites the Buddha.	King	Kṣemaṃkara	Steam bath, Incense, Monastery	Consistent	6th—II
	Outer	②	The king offers an umbrella and prays to the Buddha.	King	Tamonuda	Jewels, Music	Consistent (Differences in offering methods	4th—II
	Inner	⑤	The king offers an umbrella and prays to the Buddha.	King	Siṃha	Umbrella	Consistent	5th—II
	Outer	③	The king shaves his head and prays to the Buddha.	Merchant	Śikhin	Monastery	Inconsistent	8th—II
Rear corridor	Outer	⑥	The merchants offer their goods to the Buddha.	Merchant	Six Victors	Gold, Jewelry, Women and Monastery	Consistent (Differences in past Buddha)	9th—II
	Inner	⑨	The monk and the king offer a lamp to the Buddha.	Princess	Ratnaśikin	Oil of lamp	Consistent	1st—I
	Outer	⑦	The young brahmin offers lotus flowers and lets his hair down to allow the Buddha to pass.	Young Brahmin	Dīpaṃkara	Lotus flowers	Consistent	3rd—II
	Inner	⑩	The young brahmin prays to the Buddha while receiving the prophecy.	Young Brahmin, Uttara	kāśyapa	Resolve to leave home	Consistent	15th—III
	Outer	⑧	The king offers a Cintāmaṇi to the Buddha.	Merchant	Sunetra	Cintāmaṇi, Monastery	Consistent (Differences in status of the donor)	10th—II
Right corridor	Outer	⑪	The monk offers a robe to the Buddha.	Merchant	Vāsiṣṭha	Grove, Monastery	Inconsistent	13th—III
	Inner	⑭	The merchants pray and make offerings to the Buddha as he crosses the river by boat.	Leader of merchant	Aṅgirasa	Helping Buddha cross the river	Consistent	11th—II
	Outer	⑫	The king shaves his head. *	Merchant	Kanakamuni	Grove, resolve to leave home **	Consistent (Differences in status of the donor)	14th—III
	Inner	⑮	The brahmin offers food (or incense) to the Buddha and sees him off.	King	Manoratha	Faith	Inconsistent	12th—III
	Outer	⑬	The brahmin offers leopard-print cloth to the Buddha. *	Ṛṣi	Sunetra	Bark clothes	Consistent	2nd—I

Sources: Author’s own compilation; Le Coq (1913); Grünwedel (1924); Murakami (1984); Meng (1995); Yao (2013); 84000 Translation Project (2021). Notes: The numbers in column 2 present subjects based on Le Coq’s praṇidhi painting numbering. * Denotes the subject whose iconography no longer remains; ** denotes the subject whose inscription no longer remains; see the note to Table 1.

, uses the sequence to organize the content of the praṇidhi paintings. Table 2 shows the placement and arrangement of the praṇidhi paintings, their iconographic and inscription content, and their relationship to the Bhaiṣajyavastu. Column 1 of Table 2 indicates the location of murals; column 2 presents the sequence of the praṇidhi paintings; column 3 is the iconographic content of the murals; columns 4 through 6 highlight the inscription content according to the status of the donor, the name of the past Buddha, and the method of offering. Column 7 presents whether the inscriptions and iconography correspond to each other; column 8 shows the ordering in which the inscriptions are found in the Bhaiṣajyavastu. The numbers (1–15) in this column indicate the ordering of the 15 subjects in the Bhaiṣajyavastu, while the Roman numerals (I, II, and III) denote the asaṅkhyeya kalpas to which the subjects belong. I focus on exploring a pattern that explains how the inscription order in column 8 derives from the Bhaiṣajyavastu, based on the rearrangement of the praṇidhi painting.

To understand the extraction pattern of inscriptions from the Bhaiṣajyavastu, I first examine Subjects 9, 7, and 10 in the rear corridor, taking into account both their placement at the inner center of the corridor and their significance as key scenes within the overall iconographic program of the praṇidhi paintings. Considering a possible pairing between the inner and outer walls, when viewers enter the rear corridor, they may have viewed the murals in a zigzag sequence between both walls: Subject 9 on the inner wall, Subject 7 on the outer wall, and then Subject 10 on the inner wall (see Figure 2). As shown in column 3 of Table 1, the inscription of Subject 9 includes the expression “at the end of the first asaṅkhya [kalpa],” Subject 7 contains “at the end of the second asaṅkhya [kalpa]” and Subject 10 includes “at the end of all virtuous practices of the third asaṅkhya [kalpa]”—these expressions do not appear in the verses of the Bhaiṣajyavastu but were added at the end of these three inscriptions. This is why these subjects have importance over the entire set of praṇidhi subjects. On this basis, we should first look for where these three inscriptions are derived from the text.

Through tracing the passages in the Bhaiṣajyavastu that have the inscriptions of the three subjects, it is noticed that the inscription of Subject 9 (Figure 5), describing a princess offering oil for a lamp to Ratnaśikin, appears in the early part of the first asaṅkhya kalpa, which means this inscription is the first of the 15 praṇidhi painting inscriptions. Next, the inscription of Subject 7 (Figure 6), depicting a young brahmin offering lotus flowers to Dīpaṃkara, is the first in the second asaṅkhya kalpa, and the third of the 15 inscriptions. Finally, the inscription of Subject 10 (Figure 7), depicting the young brahmin, Uttara, resolving to leave home under Kāśyapa, appears in the final part of the third asaṅkhya kalpa and is placed last among the 15 pranidhi painting inscriptions. Together, the inscriptions in Subjects 9 (Ratnaśikin Buddha), 7 (Dīpaṃkara Buddha), and 10 (Kāśyapa Buddha), with regard to their locations in the verses of the Bhaiṣajyavastu, correspond to the first (Subject 9), third (Subject 7), and fifteenth (Subject 10) positions among the 15 inscriptions on murals. For more details, see Appendix A.

Continuing to analyze the sequence of the inscriptions, I will now explain the order in which the entirety of the inscriptions is found in verses of the Bhaiṣajyavastu, starting from the inscriptions of the subjects in the entrance of the left corridor. The inscriptions of Subjects 1 and 4, recording incense offerings and issuing invitations to Mahendra Buddha and Kṣemaṃkara Buddha; and the inscriptions of Subject 2 and 5, describing music offerings to Tamonuda Buddha and umbrella offerings to the Siṃha Buddha, are in the entrance of the left corridor. The inscriptions of these subjects are found successively following the inscription of Subject 7, in the second asaṅkhya kalpa. The order shows that the inscriptions of Subjects 2 and 5 are placed earlier in the text and that they are followed by the inscriptions of Subjects 1 and 4. Specifically, the inscriptions of Subject 2 (Tamonuda Buddha) and Subject 5 (Siṃha Buddha) have the fourth and fifth positions in the order of the 15 inscriptions, while those of Subject 4 (Kṣemaṃkara Buddha) and Subject 1 (Mahendra Buddha) have the sixth and seventh positions.

Subsequently, we now explore the order of the inscriptions of the subjects in the left-outer and rear-outer corridors. The inscriptions of Subject 3 (Śikhin Buddha), merchant’s monastery offering; Subject 6 (Six Victors), merchants’ treasure offerings;10 and Subject 8 (Sunetra Buddha), the merchant’s Cintāmaṇi offerings belong to the second asaṅkhya kalpa and correspond to the eighth, ninth, and tenth positions. In the right corridor, we have the inscriptions of Subject 14 (Aṅgirasa Buddha), the Buddha crossing the river by boat; Subject 15 (Manoratha Buddha), the king’s offering of faith; Subject 11 (Vāsiṣṭha Buddha), the merchant’s monastery offering; and Subject 12 (kanakamuni Buddha), the merchant’s grove offering and resolving to leave home. These inscriptions are found in the verses of the second and third asaṅkhya kalpas. Specifically, Subject 14 belongs to the second asaṅkhya kalpa and corresponds to the eleventh position. The inscriptions of Subjects 15, 11, and 12 are found in the third asaṅkhya kalpa, corresponding to the twelfth, thirteenth, and fourteenth positions in the 15 inscriptions (see Appendix A and Appendix B).

Approaching the exit of the corridor, we find Subjects 15 and 13. It is noteworthy that the inscription of Subject 13 (Sunetra Buddha), ṛṣi11 (seer)’s bark clothing offering, was derived from the first asaṅkhya kalpa and that it is placed as the second of the 15 inscriptions. As seen in column 8 Table 2, most of the inscriptions in the murals originate from the second and third asaṅkhya kalpas of the Bhaiṣajyavastu. Hence, the fact that the final subject (Subject 13) draws from the verses of the first asaṅkhya kalpa could symbolize the beginning of a new worship cycle. In the context of the spatial features and ritual circumambulatory pathways of Cave 20, Subjects 15 (Figure 8b) and 13 (Figure 8a) constitute the final part of the iconographic program of the praṇidhi paintings. Although Subject 15 looks similar to Subject 1, the beginning mural of the left corridor, I suggest that its iconography should not be regarded as an ‘invitation’ as expressed in Subject 1 but as a gesture of farewell to the Buddha. Through its inscription, Subject 15 states that “the king offered faith and be called Dharma King,” which may express an epilogue in the narrative cycle by emphasizing the need for cultivation of devout faith. At the same time, the inscription of the opposing Subject 13, which again draws from the first asaṅkhya kalpa, can be understood as expressing the initiation of another devotional cycle (see inscriptions in column 3 of Table 1). Taken together, the narrative sequence of the praṇidhi paintings begins in the left corridor and ends in the right corridor, with Subject 13 suggesting another new cycle of devotion.

To conclude this section, I would like to summarize my main findings about how the inscriptions are traced to the Bhaiṣajyavastu. First, the inscriptions of Subjects 9, 7, and 10 in the rear corridor were likely given prominence in selection, each drawn respectively from the three asaṅkhya kalpas of the Bhaiṣajyavastu. Subsequently, the inscriptions in the left corridor are arranged such that the subjects on the inner and outer walls correspond, appearing sequentially from Subjects 2 and 5, then Subjects 4 and 1, drawn from the second asaṅkhya kalpa. Next, inscriptions appear on Subject 3 on the outer wall of the left corridor, followed by Subjects 6 and 8 on the outer wall of the rear corridor, and then continue sequentially along the inner wall of the right corridor up to Subject 14, still derived from the second asaṅkhya kalpa. From Subject 15 on the inner wall of the right corridor, through the outer wall with Subjects 11 and 12, the inscriptions are drawn from the third asaṅkhya kalpa, while the inscription of Subject 13 on the outer wall is taken from the first asaṅkhya kalpa. Stated briefly, the sequence of the inscriptions in relation to the Bhaiṣajyavastu, inscriptions of the key scenes at the center of the rear corridor, are drawn from a well-known story of each asaṅkhya kalpa. As we proceed from the left to the right corridors, the inscriptions of the praṇidhi paintings are derived from the second to the third asaṅkhya kalpas, and finally to the first asaṅkhya kalpa.

These findings prompt two questions: first, why do the inscriptions follow the sequence from the second, to the third, and then the first asaṅkhya kalpa from the left corridor instead of a common order—first, second, and third asaṅkhya kalpa?; and second, why are the key scenes concentrated at the center of the rear corridor and how should they be understood within the broader sequence of the praṇidhi paintings?12

Starting with the first question, contrary to Le Coq’s report that no murals have survived in the antechamber,13 the antechamber of Cave 20 in fact preserves remnants of murals, with the lower part of the wall still showing the feet of nine standing Buddhas set against a natural background (Figure 9 and Appendix D).14 Although donor figures are not arranged at the bottom of the composition, the standing Buddhas are depicted wearing sandals and standing on a lotus pedestal, as in the praṇidhi paintings in the corridor. The presence of the nine standing Buddhas may suggest that the Cave 20 praṇidhi paintings form an iconographic program connecting the antechamber and the corridor, representing a total of twenty-four Buddhas. Therefore, there is a possibility that the standing Buddhas in the antechamber could be associated with the Buddhas of the first asaṅkhya kalpa. In this context, the larger iconographic program, including the antechamber, may constitute a circular structure: the Buddhas of the first asaṅkhya kalpa in the antechamber serve as the starting point, the second asaṅkhya kalpa unfolds in the left corridor, and the third asaṅkhya kalpa in the right corridor leads back once again to the first asaṅkhya kalpa. As mentioned earlier, Subject 13 at the end of the right corridor appears to have served the role of linking the iconographic programs of the corridor to the antechamber and can be understood as simultaneously completing the circumambulatory ritual while proposing a new cycle of worship. If we consider the nine Buddhas in the antechamber, which could represent the first asaṅkhya kalpa, the left corridor, the beginning of the corridor cycle, would be second in the overall sequence of murals. That is why the proposed order system starts with the second, to the third, and then the first.

Considering the second question, the three key scenes of the first, second, and third asaṅkhya kalpa in the rear corridor appear to form a central core. Similarly to how the Sarvāstivāda assigns Ratnaśikin, Dīpaṃkara, Vipaśyin, and Kāśyapa to each of the asaṅkhya kalpas and the 91 kalpa period, the praṇidhi paintings of Cave 20 appear to group the themes of Ratnaśikin, Dīpaṃkara, and Kāśyapa—representing the first, second, and third asaṅkhya kalpa, respectively—and place them at the center of the narrative sequence. Furthermore, it is worthwhile noting that these three subjects are located in the darkest space of the corridor, as determined by the architectural structure. When illuminated, they would likely have been conveyed to viewers with greater tension and visual impact, suggesting the possibility that some form of specific devotional practice took place in space.15 This visual scheme seems to condense the core thematic content of the murals into the mid-point of the corridor system. In this sense, the concentration of the three key scenes at the center of the rear corridor can be interpreted as an elaborately conceived scheme to integrate three core themes in a small inner cycle within the larger circular program of the praṇidhi paintings.

Thus far, by examining the praṇidhi murals through an alternative arrangement of murals, we demonstrated a pattern according to which the inscriptions were selected from the Bhaiṣajyavastu. As an additional issue, given that inscriptions originated from the Bhaiṣajyavastu, researchers could relate these findings to the association of the paintings with the Bhaiṣajyavastu by examining the relationship between the inscriptions and the iconographic theme of the praṇidhi paintings, particularly about what interpretive role the inscriptions may play in understanding the iconography (Murakami 1984, p. 343; Zhu 2012, p. 80; Konczak 2012, pp. 44–45; Kim 2013, p. 37). However, for the following reasons, the iconography cannot be regarded as originating directly from Bhaiṣajyavastu. As shown in column 7 in Table 2, a few of the praṇidhi painting scenes—Subject 3, Subject 11, and Subject 15—do not correspond to their inscriptions, in that neither the status of the donor nor the methods of offering depicted in the paintings align with the content of the inscriptions. These discrepancies imply that the paintings might not directly draw from the content of the inscription.16 Second, in the Turfan region, praṇidhi paintings are found in seventeen caves and temples: Bezeklik Cave, 15, 18, 20, 22, 24, 31, 33, 37, 38, 42, 47, 48, 5017 and Kocho Temple V, Temple α, Temple β and Sengim Temple No. 1 (Grünwedel 1906, pp. 41–134). Among these, the praṇidhi paintings with inscriptions have so far been identified in Bezeklik Cave 20 and Sengim Temple No. 1. In this paper, we discussed Cave 20 in detail and only briefly Sengim Temple No. 1. This means that the praṇidhi paintings with inscriptions may not have been a common convention in the Gaochang Uyghur Kingdom. Moreover, although Bezeklik Cave 15 shares the same iconographic program as Cave 20, its 15 scenes of praṇidhi paintings contain a designated cartouche that remains empty of any actual inscriptions. It is puzzling that the horizontal cartouche was rendered at the top of each scene but left blank. In that sense, we may think that inscriptions were not necessarily required for the praṇidhi paintings. Therefore, even if the inscriptions of Cave 20′s praṇidhi paintings were deliberately and carefully selected from the Bhaiṣajyavastu, which provides us with crucial information for understanding the paintings, this does not necessarily mean that the iconography was produced on the basis of the Bhaiṣajyavastu.

4. Conclusions

Through an alternative arrangement of murals based on Grünwedel’s book in 1924, this article examines the spatial placement, sequence, iconography and inscriptions of the praṇidhi paintings of Cave 20, with a particular focus on their relationships with the Bhaiṣajyavastu. To comprehend the pictorial program of the 15 praṇidhi paintings within the cave corridor, the murals should be arranged in a sequence beginning from the left corridor and exiting out of the right corridor, with the murals on the inner and outer walls of each corridor forming paired relationships.

Importantly, my research shows that the inscriptions of all subjects of the praṇidhi paintings, except for the unpreserved Subject 12, are traced to the verses of the Bhaiṣajyavastu. We saw that the inscriptions of Subject 9 (Ratnaśikin Buddha), Subject 7 (Dīpaṃkara Buddha), and Subject 10 (Kāśyapa Buddha) in the rear corridor are selected respectively from three asaṅkhya kalpas of the Bhaiṣajyavastu, each representing the pivotal scene of its respective period. Having first clarified these three key subjects, from the left toward the right corridors, I also found that the inscriptions are derived sequentially from the second to the third, and then the first asaṅkhya kalpa of the Bhaiṣajyavastu. Following this sequence, the praṇidhi paintings could illustrate a pictorial narrative: it begins with inviting the Buddha; proceeds to the king’s offering of an umbrella and worship to the Buddha; continues with making vows and receiving predictions from Ratnaśikin, Dīpaṃkara and Kāśyapa Buddha; and concludes with Buddha being seen off, perhaps signifying the revival of the viewer’s worship.

In conclusion, this article aimed to identify the difficulty in the existing literature in clarifying how the inscriptions correspond to the Bhaiṣajyavastu. Overall, by applying the idea that the praṇidhi paintings were an integral part of the cave corridor, I demonstrated that the inscriptions represent meaningful selections from the Bhaiṣajyavastu and that the iconography conveys a coherent pictorial narrative. These findings, as stated earlier by Grünwedel, may indicate that the praṇidhi paintings of Cave 20 were related to the circumambulatory religious practice conducted along the cave corridor.