"Rhapsody on the Ershi General Spring" (Ershi quan fu) from Mogaoku, Dunhuang: Introduction and Translation.

• Author: Sanft, Charles

The document trove from the library cave at Mogaoku, Dunhuang (Gansu), evokes wonder and despair. In their quantity and variety, the texts reflect the tremendous wealth of the late medieval intellectual world as it was, even in a desert oasis, far from the centers of culture. By that illumination, Mogaoku adumbrates what has been lost.

Mogaoku is most famous for its religious art and texts, especially Buddhist sutras, but the secular word is present among the manuscripts, too, representing some ten percent of the hoard, according to Imre Galambos. (1) Poems that were no longer extant--and perhaps never circulated beyond the local area--are numerous. Their topics include history and quasi-history. One such poem is "Rhapsody on the Ershi Spring" (Ershi quan fu), known from two copies (P.2488 and P.2712) and a fragment (P.2621) from the Mogaoku library cave.

"Rhapsody on the Ershi Spring" treats the famous Han dynasty (206/202 BCE-220 CE) general Li Guangli (d. 89 BCE), better known by his sobriquet, the "Ershi General" (Ershi jiangjun), here shortened to just "Ershi". In this article I expand the title of the rhapsody into "Rhapsody on the Ershi General Spring." This poem relates a legend, according to which Li Guangli created a spring to provide water for his troops, who had run out while traversing the desert reaches of the northwest border region. The present article provides an introduction, transcription, and translation of "Rhapsody on the Ershi General Spring," including details about extant manuscript copies, the question of authorship, and the content of the poem.

DUNHUANG RHAPSODIES

"Rhapsody on the Ershi General Spring" is an example of "Dunhuang literature" (Dunhuang wenxue). Most scholars use that label for the multitude of texts dating to the Tang (seventh to tenth centuries), Five Dynasties (tenth century), and early Song (late tenth to early eleventh century) periods that emerged from the Mogaoku library cave, which was sealed early in the eleventh century and reopened in 1900. (2) In this usage, Dunhuang literature is an ad hoc label for a highly varied corpus, not a generic designation. Lin Xueling has recently proposed looking also--perhaps even instead--for content that has a connection with Dunhuang. That would exclude some texts from Mogaoku and bring others under the umbrella of Dunhuang literature. Whether or not one accepts Lin's revision, it is clear that many texts from Mogaoku, especially secular texts, evince an interest in Dunhuang as a place. This sometimes takes the form of relating legends about the area and its heroes. Lin identifies "Rhapsody on the Ershi General Spring" as an example of a Dunhuang legend, which is certainly correct. I would point out that Li Guangli also counts as a hero, at least in this telling, though he is less specifically local than the legend is. The story also touches upon water, which, as Lin notes, is a preoccupation of much Dunhuang literature. (3) Given the aridity of the region, that is unsurprising.

The title of the piece, which the manuscripts preserve, leaves no doubt that "Rhapsody on the Ershi General Spring" is a fu, a "rhapsody." The fu as a poetic form developed from early origins to reach its peak during the Han period. Han rhapsodies show the poetic imagination working to its fullest in descriptions of places, events, and travel, employing rich and elaborate description and extraordinary vocabulary. Their prosody is irregular, mixing rhyme and blank verse with prose and dialogue. (4) By the Tang period, the "vernacular rhapsody" (sufu) had developed out of those antecedents. While distinct from the Han rhapsody, the medieval vernacular rhapsody also mixed prose and poetry in an irregular form. The language and presentation is, however, simpler, as befits poems created for oral performance, and include apparently colloquial formulations. Traditional scholars were acquainted with the vernacular rhapsody but granted it little value as literature. That began to change with the recovery of numerous rhapsodies from Mogaoku, which Ma Jigao and others identify as examples of the vernacular form. Cao Minggang notes that these Dunhuang fu often relate legends and historical tales, as in the case of "Rhapsody on the Ershi General Spring." Their creators seem to have tailored their topics and poetic style to entertain an audience of commoners. (5)

MANUSCRIPT RHAPSODIES FROM MOGAOKU

Pan Chonggui identifies some fifteen rhapsodies among the manuscripts from Mogaoku, which are, like much of the Dunhuang corpus, spread between Paris, London, and St. Petersburg. (6) The Bibliotheque nationale de France holds two more or less complete copies of "Rhapsody on the Ershi General Spring" (P.2488 and P.2712), along with one very short fragment of it (P.2621). The library makes high-quality images of the manuscripts available online. (7) The fragment P.2621 is too short to provide much useful information, and I concentrate here on the two extensive copies.

Imre Galambos has written about two forms of composite text that are common among the materials from Mogaoku. One comprises texts that were copied separately on multiple pieces of paper and then physically combined to create a unit. The other was also made of more than one sheet of paper, with content from various sources, but its copyist wrote the content out as an integral whole. These two kinds of composite text occur as scrolls, codices, and in other forms. (8)

Galambos's studies of composite texts to date have treated those with Buddhist content. P.2488 and P.2712 are secular examples of the latter, integral type of composite scroll. Both contain "Rhapsody on the Ershi General Spring" together with other poetry, which a copyist has written out anew. P.2488 is made up of five pieces of paper that constitute a scroll 126.4 cm long and ranging 30.1-30.5 cm in height. The writing material is "ordinary, fairly thick, fibrous beige paper," with spots of ink and glue, as well as some brown spots on one of the sheets. (9) P.2712 is made from two sheets of paper, both of which have been damaged over the centuries. Its total length is 75.7 cm and its height varies from 28.7 to 29.1 cm. The paper is of good quality, though of an irregular pulp, beige in color, with some holes and tears. The writing surface is splotched from humidity and bears spots of fat (perhaps tallow) and, on the reverse side, ink. (10) There are minor textual variations between the two copies of "Rhapsody on the Ershi General Spring," as one would expect, but they clearly record the same work.

The manuscripts bear dates of copying on their verso sides; these dates are, of course, distinct from the date of composition. P.2712 comes from, "the 6th year of Zhenming, the gengchen year, 2nd month, 19th day", which in standard reckoning corresponds to March 12, 920." While P.2488 has a date, it does not specify a reign period and reads only, "xinmao year, first month, 8th day". There were three xinmao years in the late ninth through tenth centuries and it would seem that this copy must come from one of those: 871 (which would correspond to February 1, 871), 931 (January 29, 931), and 991 (January 26, 991). (12) Scholars have suggested 871 and 931. Since the most recent date on a document from Mogaoku corresponds to 1002, 991 also seems possible, although no one has suggested it. (13)

Zheng Binglin propounds the date of 871, bringing together a few different strands of argument concerning the career of the man he identifies as the rhapsodist. Zheng believes the poem to have been written before 864 and that this copy should date to not long thereafter, thus giving his 871. (14) Victor Mair and Ikeda On separately propose the subsequent xinmao year, 931, without discussion. (15) Yan Tingliang suggests 931, as well, explaining that it would make sense for the date of this copy to be relatively close to that of P.2712. Yan furthermore argues that the date of 931 would place this copy among the bulk of rhapsodies from Mogaoku, as well as at the high point of education in Dunhuang. (16) While the question cannot be resolved at present, Yan's arguments are persuasive and I accept the dating of 931.

From the perspective of the modern reader, the two manuscript copies of the poem have different limitations and points of utility. P.2488 is the better-preserved exemplar, but its copyist, Wu Gounu, was unskilled. (17) He garbled the text in several places, rendering even the count of characters uncertain. Exclusive of the title (four characters) and attribution (seven characters), which come in a single column, there are thirteen columns of text. My count gives a total of 351 content characters, along with three phonetic glosses (two graphs each) and six ditto marks. There is confused text in columns 1, 4, and 8, which results from graphs containing altered or additional brush strokes, graphic components that are spread out more widely than is usual, or apparent copying errors.

A certain Zhang Anren wrote out P.2712 in a hand better than Wu Gounu's, but its paper has suffered more from the passage of time. (18) The rhapsody's title comes in an initial column, alone, and no author is named. The text of the poem proper spreads across seventeen columns, with just a few characters in the last; the remainder of that final column begins another poem. Because of damage and fading, a number of characters are only partially or faintly visible, and a few must be inferred on the basis of empty space, context, and comparison.

There is a seven-graph character count at the end of the poem on P.2712, which states that the total is 337 characters. I counted the content characters I could detect on the photograph of P.2712, making reference as I did so to published transcriptions and to P.2488, and arrived at a total of 354 characters. The title line has six additional characters, and I furthermore count eight ditto marks and three phonetic glosses of two graphs...