Abstract
Wang Xizhi 王羲之 (ca. 303-ca. 361), the paragon calligrapher of the Eastern Jin dynasty, became a canonized figure in Chinese cultural history, particularly after Emperor Taizong of the Tang obsessively collected and reproduced his works. At that time, thus, one main criteria for ideal calligraphy was its resemblance to Wang’s style. In this context, stele inscriptions emerged that were composed by collecting, comparing, and imitating individual characters from Wang’s extant corpus—a practice known as “Collating Characters” 集字. One notable example in Korea is the Memorial Stele for Enshrining the Amitābha Buddha Statue at Mujangsa Temple 鍪藏寺阿彌陀佛造成記碑(801). Controversy has surrounded its calligraphic origins, however. In 1803, the prominent Qing scholar Weng Fanggang 翁方綱 (1733–1818) stated that the stele’s calligraphy was modeled on the Dingwu edition (1041) of the Preface to the Orchid Pavilion. In contrast, his son Weng Shukun 翁樹崑 (1786-1815) and the Korean antiquarian Kim Chŏnghŭi 金正喜 (1786–1856) maintained the traditional view, i.e., the brushwork to Kim Yukchin 金陸珍 (fl. tenth century), a Silla calligrapher. This case study of the Mujangsa Stele examines how the same inscription was interpreted differently by scholars in China and Korea, revealing divergent frameworks of copying, authenticity, and cultural authority. It then turns to ongoing debates among modern scholars, proposing that the two seemingly opposing theories—collation versus Korean inscriber—may in fact be complementary rather than contradictory.
-
Keywords: Wang Xizhi 王羲之, Preface to the Orchid Pavilion, Preface to Sage Teachings, Mujangsa Stele, Weng Fanggang 翁方綱, Weng Shukun 翁樹崑, Kim Chŏnghŭi 金正喜, Kim Yukchin 金陸珍, Collating Characters 集字
Among Korean inscriptions praised in China, none has been held in higher esteem than this stele.
東方文獻之見稱於中國, 無如此碑.
Kim Chŏnghŭi, 1817
Introduction*
Wang Xizhi 王羲之 (ca. 303–ca. 361) and his calligraphy were introduced to Korea through diplomatic exchanges during the formation of early Silla–Tang relations.
1 In 648, during the mission of Kim Ch’unch’u 金春秋 (602–661) to the Tang court—undertaken to secure a military alliance against Koguryŏ (37? BCE–668 CE)—Emperor Taizong (r. 626-649) bestowed upon him two calligraphic works of the
Hot Spring Stele 溫湯碑 (i.e. 溫泉銘) and the
Jin Shrine Stele 晉祠碑, both composed and inscribed by the emperor himself in the style of Wang Xizhi. Roughly a century later, Wang Xizhi’s style gained a wide currency among Korean calligraphers. Kim Saeng 金生 (b. 711) was particularly celebrated for his mastery of Wang-style calligraphy. One anecdote illustrates the high technical and aesthetic regard in which Kim’s work was held and simultaneously, Chinese lack of that recognition.
During the Chongning era (1102–1106), Academician Hong Kwan 洪灌 (d. 1126) traveled to the Song dynasty capital of Bianjing (modern Kaifeng, Henan) as part of a diplomatic mission. While residing at the Guest Hall, he was visited by Hanlin Academicians Yang Qiu 楊球 and Li Ge 李革, who had been dispatched by imperial order. As the two men took up their brushes to compose on a scroll, Hong presented them with hanging scrolls of cursive and semicursive calligraphy by Kim Saeng. Their dialogue began:
2
“We did not expect to behold original works by Wang Xizhi today,” the two exclaimed in astonishment. “These are not by Wang Xizhi,” Hong replied. “They are the work of Kim Saeng, a man of Silla.” The two laughed and said, “Apart from the works of Wang Xizhi, how could such marvelous calligraphy exist in this world?” Hong explained it several times, but they did not believe it to the end.
二人大駭曰: “不圖今日得見王右軍手書.” 洪灌曰: “非是, 此乃新羅人金生所書也.” 二人笑曰: “天下除右軍, 焉有妙筆如此哉.” 洪灌屢言之, 終不信.
This exchange occurred during the early reign of Emperor Huizong (r. 1100–1125), a period marked by a revivalist zeal for antiquity and imperial enthusiasm for collecting paintings and calligraphy. Like Emperor Taizong of Tang, Huizong avidly sought works attributed to Wang Xizhi and made them copied. At a time when even the most accomplished Chinese calligrapher struggled to emulate Wang’s style convincingly, the two Chinese scholars could not accept the possibility that a non-Chinese calligrapher could have produced such work. This anecdote highlights not only the technical brilliance of Kim Saeng’s calligraphy, but also the deeply rooted Sinocentric view that artistic excellence was presumed to be coterminous with Chinese cultural identity.
However, it would be an overgeneralization to treat a single anecdote as definitive evidence of Chinese underestimation of Korean calligraphy. In fact, there were also voices of admiration for Kim Saeng. Zhao Mengfu 趙孟頫 (1254–1322), a master calligrapher of the Yuan dynasty, offers high praise for the Korean master in his “Postface to the
Ch’angnimsa Stele” 昌林寺碑跋: “[In Kim’s calligraphy] the structure and brushwork of the characters are profoundly exemplary. Even renowned Tang inscriptions cannot easily surpass it. As the old saying goes, ‘Is there any land that does not produce talent?’ Truly, it is so.”
3
Zhao’s parallel between Kim’s inscription and the masterpieces of Tang China is notable, as it places the Korean work within the highest ranks of the Chinese calligraphic canon. The adage he invokes encapsulates a cosmopolitan vision of artistic excellence—one that transcends national boundaries and challenges the notion of Chinese cultural exclusivity. Zhao’s colophon thus functions not only as an aesthetic appraisal but also as a significant acknowledgment of non-Chinese contributions to the broader tradition of East Asian calligraphy. Unfortunately, most of Kim’s calligraphy, including the Ch’angnimsa Stele, no longer survives, making further in-depth study difficult.
More significant than Kim Saeng’s calligraphy in Sino-Korean calligraphic discourse is the
Memorial Stele for Enshrining the Amitābha Buddha Statue at Mujang Temple 鍪藏寺阿彌陀佛造成記碑 (hereafter
Mujangsa Stele). Mujangsa Temple was situated north of Amgokch’on 暗谷村, in the northeastern region of Kyŏngju. Its remote location was a deliberate choice, intended to mark a clear boundary between the sacred Buddhist domain and the secular world. In 801, one year after the death of her husband, King Sosŏng (r. 799–800), Queen Kyehwa commissioned a statue of Amitābha Buddha in devotion, to ensure his rebirth in the Pure Land, and erected a commemorative stele to document its construction.
4
Chosŏn Neo-Confucian scholars turned little attention to the
Mujangsa Stele. Long forgotten and even broken into pieces, it was rediscovered in 1770 by Hong Yangho 洪良浩 (1724–1802). During his tenure as magistrate of Kyŏngju, he launched a determined search and ultimately recovered a major fragment near the temple site. The stele’s reappearance drew significant attention due to its striking resemblance to the calligraphic style of Wang Xizhi, prompting fundamental questions about the reproduction and transmission of calligraphic models across cultural and national boundaries. The inscription had traditionally been attributed to Kim Yukchin 金陸珍 (fl. 10th century),
5 but Weng Fanggang 翁方綱 (1733–1818) offered a new and controversial theory: the stele was a case of
chipcha (Ch.
jizi 集字)—that is, the characters were collated directly from Wang Xizhi’s works.
The term collating characters refers to a uniquely Sinographic practice found primarily in China and Korea. Rooted in the reverence for proper script forms and refined calligraphic styles, this practice involved modeling one’s writing on the works of ancient masters—most notably Wang Xizhi. Practitioners compiled individual characters from Wang’s extant corpus and reassembled them to transcribe an entirely new text. In instances where the desired character was missing, they would draw upon morphologically similar components from different characters, sometimes combining elements from various periods of Wang’s life—including presumed posthumous forgeries.
Interpreting calligraphic style often involves cultural implications. Previously, the
Mujangsa Stele had been considered a brushwork of Kim Yukchin or another Silla figure, and this “inscriber theory” (
sŏjasŏl 書字說) suggests not only the adoption of Wang’s style but also its internalization and creative transformation by a Silla calligrapher. Though perhaps not equal to Wang Xizhi—the revered “Sage of Calligraphy” 書聖—this calligrapher demonstrated a level of mastery comparable to Tang dynasty copiers of Wang’s script. In contrast, Weng Fanggang’s “collating characters” theory may be read as affirming the enduring influence of Chinese cultural authority. Although collated inscriptions still reflect the individuality of the person assembling the characters,
6 the act of collation inevitably entails a higher degree of imitation than original composition. Thus, debates over the script of the
Mujangsa Stele have become a key issue in the history of Korean calligraphy, closely intertwined with questions of cultural agency and the legacy of Sinocentrism in East Asian exchange.
Against this background, the present study examines how the
Mujangsa Stele’s calligraphic style was interpreted by Qing and Chosŏn literati, focusing on Weng Fanggang, Weng Shukun 翁樹崑 (1786-1815), and Kim Chŏnghŭi 金正喜 (1786–1856). While previous research has primarily discussed historical context and issues of cultural centrism,
7 this paper shifts the focus toward the theoretical and practical aspects of collating characters.
One major challenge in this discourse is that traditional scholars often fail to provide clear justifications for their interpretations. An exception is Weng Fanggang, who points to the presence of three dots as evidence for his theory; most others, however, make assertions without offering any substantial reasoning. In its final section, therefore, the paper offers a critical review of modern scholarly debates, centering on Lee Eun-Hyuk and Jung Hyun-sook. They respectively represent the conventional view of the inscriber and the new theory of collating characters. In conclusion, I propose a reconciliatory perspective, suggesting that a synthesis of the two approaches is not only possible, but also productive for understanding the calligraphic complexity and ramifications of the Mujangsa Stele.
New Approach: Weng Fanggang and the Character sung 崇
Weng Fanggang was an eminent scholar and collector of rare rubbings and stone inscriptions. In 1779, for example, he acquired the inscription from the Sarira Pagoda of Chan Master Yong 邕禪師舍利塔銘, which was written by Li Boyao 李百藥 (565-648) and inscribed by Ouyang Xun 歐陽詢 (557-641) in 631. In celebration of this rubbing, Weng named his studio “the House of Stone and Ink” (
shimo shulou 石墨書樓).
8 His antiquarian interests extended to Korea, as evidenced by his authorship of three postscripts to early Korean steles.
9 Among these, the
Mujangsa Stele attracted his greatest attention, owing to its perceived connection with the calligraphic style of Wang Xizhi. As a specialist in the
Preface to the Orchid Pavilion 蘭亭序, Weng was able to identify traces of Wang Xizhi’s influence without difficulty. Furthermore, Weng offered the controversial theory of “collating characters.”
10
The [Mujangsa] stele’s semicursive scripts come from a mixture of the Preface to the Orchid Pavilion and characters collated by the monks Huairen and Daya. Since the Xianheng (670-674) and Kaiyuan (713-741) reigns of Tang, source materials collated [by Huairen and Daya] had been influential so that people of foreign countries practiced them. The characters they used from the Preface all matched the Dingwu edition. Therefore, we know that the edition was truly carved at the time of the Tang and thereafter it spread abroad during the same period.
碑行書, 雜用右軍蘭亭及懷仁大雅所集字. 蓋自咸亨開元以來, 唐人集右軍書, 外國皆知服習而所用蘭亭字, 皆與定武本合. 乃知定武本實是唐時所刻, 因流播於當時耳.
By the early Tang, Wang Xizhi’s original works were already scattered, prompting Emperor Taizong to obsessively collect and reproduce them. When the emperor composed the
Preface to the Sage Teaching of Tripiṭaka of Great Tang 大唐三藏聖敎序 in celebration of Xuanzang’s 玄奘 (602-664) translation of Buddhist Sutras, the monk Huairen (fl. seventh century) collated characters from Wang Xizhi’s works to produce the stele inscription at Hongfusi Temple 弘福寺 in 672. Roughly fifty years later, in 721 another monk Daya (fl. eighth century) employed the same method to erect a similar stele at Xingfusi Temple 興福寺
.
More importantly, the emperor ordered Ouyang Xun to engrave the inscription of the
Preface to the Orchid Pavillion on a stele, but it disappeared during the turbulent Tang-Song transitional period. Later, sometime between 1041 and 1048, Li Xuejiu 李學究 found it at Dingwu 定武 (modern Dingzhou 定州, Hebei province)—thus it is called the Dingwu edition, but it was lost again after the fall of the Northern Song. The earliest extant rubbing of the Dingwu stele is a Song edition, now held in the National Palace Museum in Taipei (See
Figure 2). While rubbings of the Dingwu stele were initially regarded as authentic, their credibility came under scrutiny due to the proliferation of derivative copies in later periods. As such, Weng Fanggang drew attention to the
Mujangsa Stele for its stylistic resonance with the Dingwu edition and thus claimed the authenticity of the Dingwu edition through the Silla stele.
In 1803, Weng commented on the character
sung 崇, noting its stylistic affinity with the character
chong 崇 in Wang’s works: “The three dots below the mountain radical (
shan 山) in the character
chong are fully preserved.”
11 This statement is difficult to understand without background explanation. It is thus necessary to first examine, through visual comparison, the forms of the character as it appears in the
Mujangsa Stele and in Wang’s works.
The character sung appears three times in the Mujangsa Stele, and each instance is rendered somewhat differently to avoid monotony, based on the calligraphic principle of “varied forms of the same character” 同字異形. Among the three variants, the one that Weng Fanggang noted for its “three dots” 三點 appears in the second instance. The three dots are visible beneath the san 山 radical.
In the inscription of the
Preface to the Orchid Pavilion, the character
chong is relatively easy to identify; the word chongshan 崇山 appears as a later addition inserted into the upper section of the fourth column. Among the numerous extant reproductions of the
Preface, the character
chong with three dots is observed in two editions of Dingwu editions and more distinctly in the Yuquan 玉泉 edition (See
Figures 2,
3, and
4).
12 They likely informed Weng’s recognition of the
Mujangsa Stele—particularly the character
chong and its Wang-style.
Since the original copy of the
Preface to the Orchid Pavilion already disappeared during the Tang period, it is ultimately impossible to know how Wang Xizhi actually wrote the character
chong. In his
Su-Mi zhai Lanting kao 蘇米齋蘭亭考, Weng Fanggang conducted a comprehensive examination of all available editions and documented the various forms of the character
chong 宗. In the standard cursive script, the character has no dot beneath the mountain radical 山. However, depending on the edition, a single central dot, a short horizontal stroke to the right, or a set of three dots may be added beneath the radical. The character variants are illustrated as follows
.
The first three rubbings are copies of works by Yu Shinan 虞世南 (558–638), Chu Suiliang 褚遂良 (597-658), and Feng Chengsu 馮承素 (617–672), respectively, and were collected by Emperor Qianlong as the first, second, and third editions of the
Eight Pillars of the Lanting Preface (
Lanting bazhu 蘭亭八柱). The first, known as the Zhang Jinjie edition 張金界奴本, was copied during the Tianli 天曆 era (1328–1329) and features an additional dot beneath the three strokes of the mountain radical. The second, regarded as the standard cursive form, consists of only the three original strokes. The third, the Shenlong edition 神龍本, contains a short horizontal stroke instead. The fourth and final example is the Dingwu edition, which features the distinctive three-dot form. Since our primary concern is with the three-dot variant, I will focus on the Dingwu edition in comparison with the others
.
The one dot in the Tianli edition is understood as the variation of the cursive script of the character. Nishikawa suggests that this is a distinctive trait of Wang Xizhi’s style. Influenced by Wang Xizhi, Emperor Taizong added one dot as a part of the mountain radical in the character
yan 巖 in the
Inscription of Hot Spring. In the second linage of the Shenlong edition, a short horizontal stroke is not a variation of the upper dot in the character
zong 宗—which is already present as a prominent central vertical stroke. It is therefore understood as an additional and anomalous stroke of the mountain radical.
13 Nishikawa also explains that the dot is transformed into a small horizontal stroke, placed awkwardly and imbalanced—entirely detached from the three vertical lines, while the three vertical strokes forming the
shan 山 radical appear unnatural in structure.
14
In my interpretation, however, the short horizontal stroke is understood as a transformed remnant of the three-dot form. This reading is supported by the character
chong in the
Preface to Sage Teachings, where a similar horizontal stroke appears. At the beginning of his postscript to the
Preface, Lü Haihuan 呂海寰 (1843–1927) remarks: “In this edition of the
Preface, the character
chong contains three faint small dots below the mountain radical.”
15 Contrary to his comment, the actual character form clearly exhibits a single short horizontal stroke, and it suggests that the three dots might have undergo transformation into other configurations, though at times stylized or obscured. In this vein, the horizontal stroke in the Shenlong edition is also construed as a vestige of the original three-dot form.
Most interesting is the third lineage of three-dot form in the Dingwu edition and the Mujangsa Stele. Weng Fanggang believed that the Dingwu edition is more faithful to the original than the Yuquan edition, based on the three dots as a point of reference. If his conjecture is correct, a new question arises: on what basis did Wang Xizhi add the three dots to the character? The chart above may provide a clue for the origin of the three dots.
Arguably, the earliest form of the three dots traces back to the character
chong on a Qin-dynasty bronze vessel 秦豆, in which a mountain-like form appears in the lower part of the character. In the
Shuowen jiezi 說文解字, this component was repositioned to the upper part of the character. It seems to have transformed into three dots in the
Stone Classics of the Zhengshi Era 正始石經 (240-248), also known as the
Three-Script Stone Classics 三體石經 for its inclusion of ancient script 古文, clerical script 隸書, and small seal script 小篆. The three dots are inscribed inside the roof radical (
mian 宀) of the ancient script
chong. This placement differs from later editions in which the dots appear beneath the mountain radical (
shan 山), but their presence remains discernable. Given that only about a century separates the
Stone Classics (241) and Wang’s
Preface (353), it is plausible that Wang may have drawn upon the three-dot form found in the
Stone Classics. Later, Weng Fanggang’s view was underpinned by Liu Xihai 劉喜海 (1794-1852) and Ye Zhixian 葉志詵 (1779-1862). Furthermore, Ye Changchi 葉昌熾 (1849-1916) associated Korean calligraphic practice as a result of Tang militarism.
16 Unsurprisingly, Weng’s view was widespread among contemporary Korean scholars, such as Yi Sangchŏk 李尙迪 (1804-1865) and Yi Yuwŏn 李裕元 (1814-1888). It was a somewhat expected situation, given Weng’s prominent status in Chosŏn. Exceptionally, Weng Shukun and Kim Chŏnghŭi did not concede to the mainstream view.
Weng Shukun was the sixth and youngest son of Weng Fanggang. He collected rubbings of Korean steles more extensively than his father through his connections with Korean scholars. Alongside the rubbings, the Korean scholars also provided relevant historical information, such as the author’s name and the year of inscription. This kind of information was common knowledge for Korean scholars but was needed for Chinese scholars. In return, Weng imparted epigraphical methodologies to Korean colleagues. For example, rubbings should include
piaek 碑額 “the heading of the stele” and the
chan’gyŏl 殘缺 “worn-out parts of the inscribed texts” as well as the legible characters. Weng also emphasized the material form and format of steles, such as the number of characters in each line, the total number of lines of the entire inscription, and the margins around the inscription.
17 Subsequently, Korean scholars began to examine not only content but also materiality of steles. Weng Shukun’s Korean colleagues included Hong Hyŏnju 洪顯周 (1793-1865), Sim Sanggyu 沈象奎 (1766-1838), and Sin Wi 申緯 (1769-1845), while Kim Chŏnghŭi remained his favorite friend. Weng and Kim happened to be of the same age. One token of their close friendship is Weng’s epithet, Xing-Qiu 星秋 (K: Sŏng-Ch’u); the two characters form a fusion of Weng’s courtesy name, Xingyuan 星原, and Kim’s style name, Ch’usa 秋史.
In his
Su-Mi zhai Lanting kao, as previously discussed, Weng Fanggang argued that the
Mujangsa Stele was a case of collation, but Weng Shukun did not accept this theory. Instead, he upheld the traditional view that Kim Yukchin was both the author and the calligrapher of the stele. He appended a handwritten note to his father’s book,
18 as if to correct his father’s conclusion.
Weng Shukun expressed his view again in his
Haidong jinshi lingji 海東金石零記 (Miscellaneous Notes on Korean Epigraphy).
19 The entry on the
Mujangsa Stele reads (see
Figure 6).
20
The Mujangsa Stele was the brushwork of Kim Yukchin of Silla. This is also located in Kyŏngju, though only a fragmentary edition survives. The stele stands in Kyŏngju, which is none other than ancient Kyerim. Chusa [Kim Chŏnghŭi]
鍪藏寺碑 新羅金陸珍書. 此碑亦在慶州, 只此殘本而已. 碑在慶尙道慶州, 卽鷄林也. 秋史
This style of rubbing technique is excellent. Did they perhaps use Chinese ink to achieve such luster? I request that you make rubbings again, and that send me around three to five copies.
惟種拓法甚好, 此或用中國之墨, 乃得光如是耶? 乞再拓三五紙爲禱.
[Head Commentary] Postscript is in the Suoji, a single book.
21
[頭註] 跋在瑣記之一冊內.
Weng Shukun, with the assistance of Kim Chŏnghŭi, attributed the Mujangsa Stele to the calligraphy of Kim Yukchin. He also appended a note requesting additional copies of the rubbing from Kim Chŏnghŭi for further research. However, by that time, the stele had disappeared, making it impossible to produce a new rubbing.
In addition, Weng Shukun produced a stele illustration (
pido 碑圖) based on the inscription fragments, and noted before the first line of the text, “It seems there was an additional line here” 此處似尙有一行.
22 Amid this ongoing exchange of views, Weng Shukun suddenly passed away. Upon hearing the news, Kim Chŏnghŭi was deeply grieved and, in memory of his late friend, traveled to Kyŏngju.
The site of Mujangsa Temple was overgrown by bushes when Kim Chŏnghŭi reached on the twenty-ninth day of the fourth month of 1817. Kim searched through the bushes and found the right-side part of the stele, the major fragment that Hong Yangho first found. Kim explored further and discovered another fragment, this time the left-side of the stele, which surprisingly contained the line Weng had hypothesized. Kim’s discovery proved Weng’s speculation. Kim moved the two fragments to a safe place behind the temple. Then he inscribed his own remarks on the lateral side of each stone fragment. The one written on the major one is more detailed and comprehensive.
23
This stele was formerly known only by a single fragment. During my exhaustive search, I discovered an additional broken piece amidst the overgrowth. I was overwhelmed with joy and shouted aloud in amazement. I placed the two stones together, like linked pearls, and moved them to the rear corridor of the temple to protect them from the elements. The quality of the calligraphy on this stone surpasses even that of the Paengwŏl Stele. The character sung with three dots, as in the Orchid Pavilion, is fully preserved only on this stone. Master Weng Tanxi [Fanggang] used this stele as evidence [for his theory of collation]. Among Korean inscriptions praised in China, none has been held in higher esteem than this stele. I caressed the broken stele multiple times and regretted that Xingyuan [Shukun] could not view the lower portion [that he mentioned.]
Inscribed by Kim Chŏnghŭi on the twenty-ninth day of the fourth month in the year of Chŏngch’uk (1817)
此碑舊只一段而已, 余來此窮搜, 又得斷石一段於荒莽中, 不勝驚喜絶叫也. 仍使兩石, 合壁珠聯, 移置寺之後廊, 俾免風雨. 此石書品, 當在白月碑上. 蘭亭之崇字三點, 唯此石特全. 翁覃溪先生以此碑爲證. 東方文獻之見稱於中國, 無如此碑. 余摩挲三復, 有感於星原之無以見下段也. 丁丑四月卄九日, 金正喜題識.
Kim first explains how he discovered the two broken parts of the stele and proceeds to evaluate their calligraphic merit in comparison with the Paekwŏlsŏun Pagoda Stele 白月栖雲塔碑, erected in 954 in memory of the Buddhist master Nanggong 朗空 (832–916). That inscription was engraved based on the calligraphy of Kim Saeng, the renowned Korean calligrapher introduced at the beginning of this article. Yet, in Kim’s assessment, the Mujangsa Stele still surpasses the Pagoda Stele in artistic value—not only because it predates it, but also because it preserves authentic traces of Wang Xizhi’s brushwork, such as the distinctive three-dot form of the character sung 崇, a feature that had drawn the attention of Weng Fanggang. Notably, though, Kim does not refer to Weng’s theory of “collating characters.” This is another indirect evidence that Kim did not support his master’s view.
After about fifteen-day trip to Kyŏngju, Kim Chŏnghŭi returned to Seoul and engaged more earnestly in epigraphic studies in collaboration with fellow scholars. In the course of this research, he once again expressed his views on the
Mujangsa Stele in a letter to Kim Kyŏngyŏn 金敬淵 (1778–1820). Although the precise date of the letter is not recorded, it is presumed to have been written around the autumn of 1817, a period when the two men were deeply immersed in the study of epigraphy.
24
The Mujangsa Stele is indeed written in the calligraphic style of the Hongboksa Stele, but it is not a collated inscription like the Ingaksa Stele. Kim Yukchin was a figure of the late Silla period. The date of the stele’s erection cannot be verified at present.
鍪藏碑果是弘福字體, 非集字如麟角碑矣. 金陸珍是新羅末葉之人, 而碑之年代, 今不可考矣.
Kim’s remarks are terse and ambiguously phrased, making them difficult to understand with precision. He initially affirms Weng Fanggang’s view that the inscription was rendered in the style of the
Hongfusi Stele that engraved the Preface to the Sage Teachings 聖教序. What follows, however, is a problematic passage that lends itself to two possible interpretations. One possibility is that the stele is not a collated work, unlike the
Ingaksa Stele.
25 Alternatively, the stele is a collated work, akin to the
Hongfusi Stele, but of a different kind than the
Ingaksa Stele. In my view, Kim likely intended the first interpretation, especially given his remarks in the aforementioned book Miscellaneous Notes by Weng Shukun. In the next, Kim refers to Kim Yukchin, seemingly suggesting that he was the actual inscriber, though he does not elaborate on this point. It is possible that Kim was reluctant to openly contradict the position of his mentor, Weng Fanggang.
Debates Among Modern Scholars
To this day, scholarly consensus has not been reached regarding the calligraphic style of the
Mujangsa Stele.
26 Lee Jong-moon is the earliest scholar who suggested a Korean calligrapher.
27 Yi points out that about one-quarter of the characters on the stele are not found in any extant works by Wang Xizhi. Even some characters are too complicated to assemble parts from Wang’s corpus. Yet, they still achieve the uniformity in the size and stroke breath of the characters, the lively rhythm and continuity of brush energy, and the overall harmonious composition. Those features attest that the inscription of the stele could not be made through assembling pre-existing characters.
Choi Young-sung supported Yi’s position and proposed a new theory that the calligrapher was a monk from Hwangnyongsa Temple. He deciphered the character
sa 寺 after
Hwangryong 皇龍 near the lower end of the first column on the stele, and suggested that it would have been followed by the name of the temple monk who wrote the inscription.
28
Building on the arguments of Lee Jong-moon and Choi Young-sung, Lee Eun-Hyuk further developed the debate by organizing his analysis around three aspects: character size, historical context, and calligraphic style. His main arguments can be summarized as follows. Lee first draws attention to the size of the characters on the
Mujangsa Stele in comparison to three other steles made through the collation method (see
Figure 7).
Lee Eun-Hyuk effectively illustrated the irregular distribution of character sizes in the two Chinese steles by overlaying horizontal and vertical lines on the inscriptions. The Hongfusi Stele, or Stele of the Preface to Sage Teachings, has slightly narrower spacing between characters than the Broken Stele of Xingfusi. Yet in both cases, variation in character size is clearly noticeable. In contrast, the Mujangsa Stele exhibits a high degree of formal consistency. Vertical guidelines were drawn on the stone surface to ensure even spacing between columns, and then characters were inscribed at regular intervals along these lines. The spacing between columns is orderly, and the characters are rendered in a uniform size throughout, intending a deliberate structural cohesion. Based on these observations, Yi concludes that the Mujangsa Stele was executed by a single calligrapher.
Yi further states that the Ingaksa Stele displays an uneven format, resembling the layout of the two Chinese steles. However, from my own observation, the Ingaksa Stele is still relatively consistent in character size and spacing, closely approximating the organization seen in the Mujangsa Stele. Thus, it seems reductive to interpret varying sizes solely through the lens of the collation process. A more plausible explanation would be regional stylistic conventions. Whereas Chinese steles have texts in open space, Korean ones employed vertical guidelines. The two different formats suggest disparate underlying aesthetics pertaining to spatial organization.
Meanwhile, one interpretative issue arises regarding character size. Lee Eun-Hyuk initially attributes the irregularity in character size to the inherent limitations of collation process. Because characters are collected from a variety of sources, their sizes naturally vary. Later, however, he interprets the same phenomenon as an embodiment of the so-called
canchaimei 參差美 (K.
ch’amch’imi), a key aesthetic principle in Chinese calligraphy that esteems unevenness, asymmetry, and variation.
29 In other words, it remains unclear whether the variation in character size is regarded as a technical limitation resulting from the compilation process or as a deliberate aesthetic choice. If the former is the case, it rightly points to a structural weakness inherent in collated characters. But if it is the latter, it bears no direct relevance to the issue of “collation theory.”
The second point of discussion pertains to the historical circumstance regarding the construction of the Mujangsa Stele. As evidenced by the fact that the Stele of the Preface to Sage Teachings took twenty-five years to complete, the process of collating characters requires extensive source materials and a great deal of time. In the case of the Mujangsa Stele, however, it was created as part of a Buddhist ritual intended to pray for the repose of the deceased king’s soul. Given the urgency of such a task, the time frame must have been tight—indeed, the stele was completed within a year of the king’s passing. Under such time constraints, making new characters based on collation process is highly impractical.
The second argument is overall reasonable, but it still warrants further elaboration. Lee’s argument relies solely on the example of the
Preface to Sage Teachings, making the comparison overly dependent on a single precedent. Incorporating a broader range of examples would strengthen his argument.
30 The
Preface to Sage Teachings was the first successfully executed collated stele, and thus its production likely required an exceptionally long period due to the novelty and complexity of the process. By contrast, when the
Mujangsa Stele was erected, the
Preface to Sage Teachings and other Chinese sources were available, potentially streamlining the production process and significantly reducing the time required. Moreover, the
Preface to Sage Teachings consists of 1,904 characters, considerably more than the estimated 1,400 characters of
the Mujangsa Stele.
Lastly and most importantly, Lee Eun-Hyuk classified all 431 deciphered characters from the
Mujangsa Stele into three categories based on their degree of resemblance to Wang Xizhi’s calligraphy.
31 The results of his classification are as follows:
(1) Characters identical to those in Wang Xizhi’s calligraphy: 156 characters
(2) Characters that appear in Wang’s corpus but differ in form: 161 characters
(3) Characters not found in Wang’s calligraphy: 114 characters
(4) Others: Characters that are undecipherable or require further clarification
Lee’s study is grounded in direct visual comparison with Wang Xizhi’s calligraphy, and his systematic analysis provides a solid foundation for addressing the issue of collated characters. Categories (2) and (3)—that is, characters whose forms differ from those in Wang Xizhi’s works or are not found in his calligraphy—together comprise approximately 270 characters, nearly twice the number of those matching Wang’s calligraphy in Category (1). Although Category Two may vary, depending on the viewer’s judgment, the high number of characters either absent from or significantly different in Wang’s extant works supports the argument that the Mujangsa Stele is not a product of assembling characters from preexisting works. Taken as a whole, Lee Eun-Hyuk’s reasoning and progression toward his conclusion are logically sound and convincing although the two minor issues—character size and historical circumstance—need further consideration.
Among scholars in support of a single inscriber, opinions remain divided regarding the identity of the calligrapher. While Choi Young-sung and Lee Eun-Hyuk suggest a Hwangnyongsa monk or a third party, Lee Jong-moon continues to regard Kim Yukchin as the calligrapher, based on textual evidence.
32 Since this study focuses primarily on the broader debate between the single-inscriber and collation theories, internal disagreements within the former camp will not be addressed in detail here.
Now let us turn to the arguments of scholars advocating the collation theory. In this group of scholars,
33 Kim Ŭnghyŏn is the first scholar to advance a scholarly argument of the collation theory. He explained that even collated inscriptions can reflect the individuality of the person assembling the characters.
34 Building upon his discussion, Jung Hyun-sook developed the theory further.
35 The script style of the
Mujangsa Stele differs from that of Wang Xizhi in the Chinese sources, but such a level of difference is not significant to assert a single inscriber on the ground of the process of collating characters. The process first involves locating the relevant characters, which are then manually copied one by one; as such, the final form may vary depending on the collator’s skill and stylistic choices. In her view, therefore, the scriptural differences between the
Mujangsa Stele and the works of Wang Xizhi do not necessarily support the theory of a single inscriber.
Moreover, Jung Hyun-sook draws attention to the formal aspects of the inscription. While she accepts Choi Young-sung’s claim that the name of a monk was engraved at the bottom of the first line of the stele, she argues that this monk was not the inscriber but rather the monk in charge of collation. As evidence, she pointed to the fact that in both the
Preface to Sage Teachings and the
Broken Stele of Xingfusi Temple, the name of the collating monk follows the temple name at the beginning of the inscription.
36
The format of the Preface to Sage Teachings is particularly noteworthy; a large blank space is between the phrase “Composed by Emperor Taizong” 太宗文皇帝製 and “Hongfusi Temple” 弘福寺. This format aligns with that of the Mujangsa Stele, where “Taenaema Minister Kim Yukjin, in obedience to the request” 大奈麻臣金陸珍奉敎 is followed by a blank of more than four characters before “Hwangnyongsa” 皇龍寺 is written. Based on such a similar format, Jung infers that the full inscription of the Mujangsa Stele likely reads: “Hwangnyongsa [‘a name of monk’ collated writings of General Wang Xizhi of Jin]” 皇龍寺 [□□ 集晉右將軍王羲之書].
Jung Hyun-sook further notes that when a monk serves as an inscriber, there is no precedent for his name to appear at the beginning of an inscription. By contrast, in the case of collated works, the collator’s name is typically placed at the beginning, as seen in the two aforementioned Chinese steles and in Silla’s
Master Honggak Stele (886). On this basis, Jung concludes that the Hwangnyongsa monk in question should be regarded not as an inscriber, but as a collating monk. On the whole, Jung articulates a well-reasoned argument, grounding her analysis in the concept of collation and formal comparisons with other compiled steles. However, we need to re-consider her claim that the name of the collating monk appears at the beginning of the inscription. One counter example is the
Ingaksa Stele, whose inscription ends with:
Disciple-monk Jukhŏ, having received the imperial edict, collated the characters of General Wang Xizhi of Jin. Disciple Ch’ŏngbun, Abbot of Palace Temple and Ingaksa Temple, T’ong’o Chinjŏng Grand Sŏn Master erected the stele.
門人沙門竹虛奉 勅集晉右將軍王羲之書. 門人內願堂兼住持通奧眞靜太禪師淸玢立石.
As shown above, the name of the collating monk appears at the very end of the inscription. It suggests that a more comprehensive investigation is necessary for validating Chŏng’s argument.
Concluding Remarks
I have reviewed modern scholarly works with a main focus on Lee Eun-Hyuk and Jung Hyun-sook, followed by my own view. First, Lee and Jung stand at the opposite sides, but they have shared views at times. Both scholars noted that the script of the Mujangsa Stele most closely resembles that of the Hongfusi Stele, i.e., the Stele of the Preface to Sage Teachings. This observation aligns well with the historical context. At the time the Mujangsa Stele was erected, the Preface to the Orchid Gathering was already an extremely rare text and difficult to obtain, while the Broken Stele of Xingfusi had not yet been discovered. Thus, the Hongfusi Stele would have been the most accessible and useful source for reproducing the calligraphic style of Wang Xizhi.
As for the script style of the Mujangsa Stele, both Lee and Jung exhibit some points of convergence in their analysis, but they differ in interpretation and emphasis. Attention to these differences reveals the core of the debate and may offer insights toward its resolution. Lee remarks that the strokes are uniform and sharp. Similarly, Jung observes that, compared to the two Tang-dynasty steles, the strokes are thinner and exhibit characteristics of regular script, producing a “lean and rigid” (sugyŏng 瘦硬) feature. As such, their descriptive accounts largely coincide, but their interpretations diverge. Lee argues that the evenness of the brushwork and the regular spacing between characters point to a single calligrapher’s hand. In contrast, Jung maintains that since collation involves a manual process of copying at the end, such a consistent script style happens naturally.
Now I would like to offer my own view. While both Lee and Jung make valid points on their own, each appears to have limitations when it comes to evaluating the script of the
Mujangsa Stele in its entirety. Lee notes that 156 out of the 431 deciphered characters in the stele match Wang Xizhi’s calligraphy. Although this number does not account for half, it still represents a significant portion. These characters were likely compiled based on extant materials such as the
Preface to Sage Teachings. Jung Hyun-sook, on the other hand, argues that for characters not found in Wang’s corpus, the collating monk would have had no choice but to write them in the style of Wang Xizhi.
37 This statement effectively acknowledges the role of the inscriber to a certain degree. Given that a substantial number of characters in the stele are not attested in Wang’s model calligraphy, they cannot be dismissed as negligible.
Considering that both theories have some limitations, it seems most reasonable to adopt a hybrid view: characters traceable to Wang’s models were collated, while those absent were written directly by the inscriber. Song Mingxin previously proposed such a view that both collation and freehand copy are evident in the inscription. He concluded that the script of the
Mujangsa Stele represents a creative adaptation of Wang Xizhi’s calligraphy.
38
The most contentious issue concerns a group of Chinese characters that appear in Wang Xizhi’s calligraphy but take different forms in the Mujangsa Stele. Lee identifies a total of 161 such characters and, based on this observation, concludes that the inscription reflects one individual’s creative handwriting. In contrast, Jung attributes the variations to the collating monk’s level of skill and personal style. The two opposite interpretations may involve subjective judgment, and the issue exceeds the present author’s capacity for definitive resolution. Thus, rather than pursue the issue further here, I leave it open to future research.
Lastly, let us return to the varied forms of the character
sung 崇. Thus far, scholarly discussions have focused on the presence of the three dots. However, equal attention should be given to the lower component 宗 (K.
chong, Ch.
zong). In the
Mujangsa Stele, this component appears in its standard form, whereas in the Dingwu editions, the thick central stroke of the character pierces straight through the entire
zong component, as Weng Fanggang mentioned (see the chart above).
39 In other words, the Korean inscriber adopted the three-dot form but did not replicate the penetrating stroke, which is also characteristic of the Dingwu editions. This divergence highlights the creative agency of the Korean calligrapher, whose approach reflects a selective and interpretive engagement with the model rather than mechanical reproduction.
Notes
Figure 1.Details of rubbings of the Mujangsa Stele. Dimensions unknown. National Research Institute of Cultural Heritage.
Figure 2.Beginning section of the Dingwu edition. Handscroll, dimensions unknown. National Palace Museum in Taipei. Image number: K2D000001N000000000PAC.
Figure 3.Beginning section of the Dingwu edition. Handscroll, Height 24 cm, Width 9.5 cm. Image number: 宋拓定武兰亭序 新00135367. The Palace Museum, Beijing.
Figure 4.Beginning section of the Yuquan edition. Dimensions: Height 25.2 cm, Width 11.1 cm, Kyoto National Museum, Japan. After Ō Kishi to Ranteisho 王羲之と蘭亭序, 27.
Figure 5.
Figure 6.
Haidong jinshi lingji 海東金石零記, Entry. Mujangsa Stele
Figure 7.From left to right: (a) Huairen, Hongfusi Stele (672), (b) Daya, Broken Stele of Xingfusi (721), (c) Kim Yukjin (arguably), Mujangsa Stele (801), and (d) Chukhŏ, Ingaksa Stele (1295). From Lee Eun-Hyuk, “Mujangsa pi wa Wang Hŭiji ch’e ŭi taebi koch’al,” 186.
Figure 8.Rubbing of the Ingaksa Stele Inscription, 1295. Accordion-fold Album (12 Folds), 39.2 × 27 cm. Jangseogak no. B14B 29
Chart: The Graphical Development and Variants of the Character Chong
REFERENCES
- Choi, Young-sung. ““Silla Mujangsa pi ŭi sŏja yŏngu” 新羅 鍪藏寺碑의 書者 硏究 [A Study on the Calligrapher of the Mujangsa Stele of Silla)].” Sillasa hakpo 20 (2010): 179-218.
- De Laurentis, Pietro. Protecting the Dharma through Calligraphy in Tang China: A Study of the Ji Wang Shengjiao Xu: The Preface to the Buddhist Scriptures Engraved on Stone in Wang Xizhi’s Collated Characters. Abingdon, Oxon: Routledge, 2021.
- Fujitsuka Chikashi 藤塚鄰. Shinchō bunka tōden no kenkyū: Kakai Dōkō gakudan to Richō no Kin Gendō 清朝文化東傳の研究: 嘉慶・道光學壇と李朝の金阮堂 [A Study of the Culture of Qing and its Reception in the East: Schools of Jiaqing and Daoguang Reigns and Kim Wandang of Chosŏn]. Tōkyō: Kokusho kankōkai 国書刋行会, 1975.
- Hong Yangho 洪良浩 (1724-1802). Igye chip 耳溪集 [Collected Works of Hong Yangho]. Han’guk munjip ch’onggan 韓國文集叢刊. Vol. 241. Seoul: Kyŏngin munhwasa, 2000.
- Iryŏn 一然 (1206-1289). Samguk yusa 三國遺事 [Memorabilia of the Three Kingdoms], Kyujanggak edition (1512).
- Jung Hyun-sook. T’ongil Silla ŭi sŏye 통일신라의 서예 [The Calligraphy of Unified Silla]. Seoul: Daunsaem, 2022.
- Jung, Hyun-sook. ““Han’guk sŏyesa esŏ Wang Huiji chipchabi ŭi ch'urhyŏn gwa chŏn’gae” 한국 서예사에서 왕희지 집자비의 출현과 전개 [The Appearance and Development of the Stelae Written in Wang Xizhi’s Characters in the History of Korean Calligraphy].” Sŏyehak yŏn’gu 44 (2024): 5-36.
- Kim Pusik 金富軾 (1075-1151), comp. Samguk sagi 三國史記 [History of Three States]. Chŏngdŏk edition (1512).
- Kim Chŏnghŭi 金正喜 (1786-1856). Wandang chip 阮堂集 [Collected Works of Kim Chŏnghŭi]. Han’guk munjip ch’onggan 韓國文集叢刊. Vol. 283-4. Seoul: Kyŏngin munhwasa, 1988.
- Kim Ŭnghyŏn. Sŏ yŏ ki in 書如其人 [Writing Mirrors the Man]. Seoul: Tongbang yŏnsŏhoe, 1995.
- Ledderose, Lothar. Mi Fu and the Classical Tradition of Chinese Calligraphy. Princeton, NJ: Princeton University Press, 1979.
- Lee, Eun-Hyuk. ““Mujangsa pi wa Wang Hǔiji ch’e ǔi taebi koch’al” 鍪藏寺碑와 王羲之體의 對比考察 [A Comparative Study of the Mujangsa Stele and Wang Xizhi’s Style].” Hanguk chǒt’ong munhwa yǒn’gu 12 (2013): 171-212.
- Lee, Jong-moon. ““Mujangsabi rŭl ssŭn sŏyega e kwanhan han koch'al” 鍪藏寺碑를 쓴 書藝家에 관한 한 考察 [A Study on the Calligrapher of the Mujangsa Stele].” Nammyeonghak yŏn’gu 13 (2002): 223-54.
- Lee, Jong-moon. ““Mujangsabi rŭl ssŭn sŏyega e taehan chae kŏmt’o” 鍪藏寺碑를 쓴 書藝家에 대한 再檢討 [A Reexamination of the Calligrapher of the Mujangsa Stele].” Taedong Hanmunhak 41 (2014): 271-302.
- Little, Stephen and Virginia Moon, eds. Beyond Line: The Art of Korean Writing. California: Los Angeles County Museum of Art, 2019.
- Liu Xu 劉昫 (888-947), comp. Jiu Tangshu 舊唐書 [Old History of Tang]. Beijing, Zhonghua shuju, 1975.
- Nishikawa Yasushi 西川寧. “Chō Kinkai do hon ni tsuite” 張金界奴本について [Regarding the Zhang Jinjie edition]. In Shōwa Rantei kinen ten 昭和蘭亭記念展 [Shōwa Exhibition Catalogue of Rantei]. Tōkyō: Nigensha, 1973, 227-238.
- Pak Chulsang [Pak Ch’ŏlsang]. Na nŭn yetkŏt i choa ttaeron kkaejin pittol ŭl ch'aja tanyŏtta: Ch'usa Kim Chŏng-hŭi ŭi kŭmsŏkhak 나는 옛것이 좋아 때론 깨진 빗돌을 찾아다녔다: 秋史 金正喜의 金石學 [I am Fond of the Ancient, And So I Searched for Broken Steles: Epigraphy of Ch’usa Kim Chŏnghŭi]. Seoul: Nŏmŏ puksŭ, 2015.
- Park, Hyun-Kyu. “[Pak Hyŏn’gyu]. “Haidong jinshi lingji ŭi sŏja wa shilsang,” 海東金石零記의 著者와 實狀 [The Authorship and Nature of Haidong jinshi lingji].” Taedong Hanmunhak 35 (2011): 385-413.
- Shin, Jeongsoo. “Kim Chŏnghŭi and His Epigraphic Studies: Two Silla Steles and Their Rubbings.” Journal of Korean Studies 27, no.2 October월 (2022): 199-223.
- Song Mingxin 宋明信. “Moucangsi bei de shuxie zhe yu shuti fenxi” 鍪藏寺碑的書寫者與書體分析” [An Analysis of the Inscriber and Script Style of the Mujangsa Stele]. Proceedings of the International Conference on the Mujangsa Stele of Silla, 2010, pp. 227-231.
- Taitō Kuritsu. Ō Kishi to Ranteisho 王羲之と蘭亭序. Shodō Hakubutsukan台東区立書道博物館 Tokyo: Taitōku gakujutsu bunka zaidan 台東区學術文化財團, 2023.
- Weng Fanggang 翁方綱. Su-Mi zhai Lanting kao 蘇米齋蘭亭考 [Investigation on the Orchid Pavilion at the Studio of Su-Mi]. In Wu Chongyao 伍崇曜 et al. Yueyatang congshu 粵雅堂叢書vol. 15. Taipei: Huawen shuju, 1965.
- ____. Fuchuzhai wen ji 復初齋文集. Beijing: Beijing University, 2023.
- Weng Shukun 翁樹崑. Haidong jinshi lingji 海東金石零記 [Miscellaneous Notes on Korean Epigraphy]. Kwach’ŏn: Weng Gwacheon Cultural Center, 2010.
- Ye Changchi 葉昌熾. Yu shi 語石. Taiwan shangwu yinshuguan 臺灣商務印書館, 1968.
- Yi Haeng 李荇 (1478-1534). Sinjŭng Tongguk yŏji sŭngnam 新增東國輿地勝覧 [Revised and Expanded Edition of Survey of the Geography of Chosŏn], Kyujanggak edition.
- Yi, P’ungmo. ““A Study of the Writings of Weng Fanggang” 翁方綱著述考.” Bibliography Quarterly 書目季刊 8, no.3 (1974): 39-45.
- Yi U 李俁. Taedong kŭmsŏksŏ 大東金石書 [Sourcebook of Metal and Stone in the Great East]. 1668.
- Comparison of transcriptions of Mujangsa Stele at the National Institute of Korean History database. https://db.history.go.kr/ancient/level.do.
- Comparison of the Four Copies of the Lanting Preface. https://patricksiu.org/four-imitation-copies-of-lanting-xu-蘭亭序摹本四種）/.
Figure & Data
Citations
Citations to this article as recorded by
News
