Textual Differences Encountered in Reading the Classics
Tang readers studying the classics faced differences among manuscripts. Similar-looking characters were easily confused, while prolonged copying produced omissions and variant readings. Teachers using different base texts might present the same passage in different forms. Because the classics were also connected with schooling and the mingjing examinations for selecting officials, deciding how the text should read and which copies to use for collation became a matter for official education.
Zheng Tan advised Emperor Wenzong to assemble scholars versed in classical learning to collate the texts and preserve the edited versions in stone, following the Later Han precedent of inscribing classics at the Imperial Academy. Zheng Tan’s biography in the Old Book of Tang records both the proposal and its acceptance. Inscription fixed selected readings in a physical object that could be checked repeatedly. Paper copies continued to serve portable reading and recitation, while uncertain characters could be compared with an officially established stone text.
References: [1]
The Division of Work: Collation, Writing, and Carving
Between the proposal and the completed monuments, the project passed through collation, preparation of written models, checking, and engraving. Zheng Tan recommended Zhou Chi, Cui Qiu, Zhang Cizong, Kong Wenye, and others to collate the classics. Surviving inscriptions name distinct responsibilities, including writing for the stones, checking the transfer of writing onto stone, reviewing character forms, and inspection. Ai Juhui, Chen Jie, and others undertook the writing, while Tang Xuandu participated in reviewing character forms. Scholars responsible for classical interpretation and wording, calligraphers, and stonecutters each had their roles in the same project.
The Stone Classics were completed in the second year of Kaicheng, 837, and erected at the Imperial Academy in Chang’an’s Wuben Ward. About 650,000 characters were carved on both sides of 114 stone slabs, forming continuous stone walls. The text was chiefly in orderly regular script, with columns and lines arranged for line-by-line consultation. Transferring such a long text to stone required every stroke to pass through writing and engraving, making the verification of forms and inspection of the finished work equally important. The completion date also gave the inscriptions their later name, the Kaicheng Stone Classics.
Twelve Classics and Appendices on Characters
The inscriptions include the Zhouyi, Shangshu, and Mao recension of the Poetry; the three ritual works Zhouli, Yili, and Liji; the three Spring and Autumn commentaries, Zuozhuan, Gongyang zhuan, and Guliang zhuan; and the Xiaojing, Lunyu, and Erya. The first nine and latter three make twelve works altogether, covering divination, government, poetry, ritual, history, and linguistic explanation. Although historical accounts retained the name “Nine Classics on Stone Walls,” the actual corpus was broader. The Mengzi carved during the Qing dynasty belongs to a later layer of supplementation.
Alongside the classics, Zhang Can’s Wujing wenzi and Tang Xuandu’s Jiujing ziyang were inscribed as appendices. Tang Xuandu explains that earlier guides to character forms had become distorted through copying, while some doubtful characters in the classics were absent from those guides. He therefore worked with the collators, consulting character dictionaries to select appropriate forms. The appendices gather easily confused forms together, allowing readers not only to establish the wording of a passage but also to compare strokes, components, and differences inherited from changes between script styles.
How Stone Texts, Rubbings, and Manuscripts Worked Together
The stone walls served practical acts of reading and collation. Readers visiting the Imperial Academy could locate passages by the title of the classic, compare doubtful readings in their manuscripts line by line, and transfer corrections to paper. Hand copying suited the extraction of particular passages, while rubbings transferred the stone lettering to paper, allowing character forms and arrangements to circulate beyond their original location. Heavy stone slabs and portable sheets thus performed complementary functions.
The stone text provided a common reference but could still contain problems introduced through writing, carving, or selection of the base text. In his Jinshi wenzi ji, the Qing scholar Gu Yanwu recorded variant characters, omissions, and miscut text by classic and compared them with other transmitted versions. For later readers, these differences reveal how Tang scholars established character forms and how texts subsequently changed. A rubbing preserves the stone’s surface at the time it was taken; a manuscript preserves a reader’s choices. Examining both alongside the original stone permits a more detailed account of a character’s transmission.
Relocation, Repair, and the Preservation of Textual Witnesses
After Chang’an was reduced in size in the late Tang period, the preservation environment of the original Imperial Academy and its Stone Classics changed, and the slabs were moved. By the Northern Song, some stood in low-lying ground threatened by accumulated water and damage. In 1087 Lü Dazhong oversaw the transfer of the Stone Classics and other steles to the north side of the prefectural school, in the area of today’s Xi’an Forest of Steles. Li Chi’s later account of the relocation described the former site and the reasons for moving, leaving a verifiable written record of the preservation process itself.
After relocation, the slabs still required repair, straightening, and replacement of missing portions. Later recutting survives alongside original carving, and researchers distinguish these layers through character forms, inscriptions, and rubbings from different periods. The facsimile Kaicheng shijing published in 2014 reproduces fine Republican-period rubbings in the museum’s collection, enabling comparison of lettering that once required on-site viewing. The official educational project completed in 837 thus left two continuing records: the form in which the classics were collated in the Tang period, and the traces of their preservation by generations of readers, stonecutters, and custodians.