Sān jiā Shī shí yí 三家詩拾遺
Recovered Fragments of the Three Schools’ Poetry by 范家相 (Fàn Jiāxiāng, zì Héngzhōu 蘅洲, d. 1769)
About the work
A 10-juǎn mid-Qiánlóng-period systematic Sān jiā Shī (Three-Schools’ Shī: Lǔ Shī, Qí Shī, Hán Shī) recovery work by Fàn Jiāxiāng — completed in Qiánlóng 25 gēngchén (1760) per Fàn’s own self-preface dated cháng zhì (Summer Solstice) plus ten days. The catalog meta’s date of 1759 represents the start of compilation; the self-preface dates the completed work to 1760.
Methodological context: by the Hàn period, four Shī schools were transmitted — Máo, Lǔ, Qí, Hán. Of these, Lǔ Shī was lost in the Xī Jìn (Western Jìn); Qí Shī in the Wèi; only Hán Shī survived through the Sòng — its Tàipíng yùlǎn extracts, Chóngwén zǒngmù listings, and Liú Ānshì / Cháo Yuèzhī’s casual citations indicate that the Hán Shī nèi zhuàn survived to the Zhènghé / Jiànyán transition (early 12th century) before being lost; the Hán Shī wài zhuàn alone (the present KR1c0066) survived. By the late-Sòng Xiánchún period (1265–1274), Wáng Yìnglín 王應麟 — under the conditions of the Mongol-Sòng war — pioneered systematic shí yí (fragment-recovery) of the lost Sān jiā Shī readings in his Shī kǎo 詩攷 (3 juǎn).
Fàn’s Sān jiā Shī shí yí is the principal mid-Qiánlóng-era systematic expansion of Wáng Yìnglín’s pioneering work. He records that he expanded Wáng’s coverage by 60–70%; he reorganized the structure: where Wáng had separated the three schools (Lǔ, Qí, Hán) into separate piān, Fàn arranged the recovered fragments by the 305 Máo Shī piān sequence, with the Sān jiā readings bundled together for each piān — facilitating cross-reference. The work’s structure (per the work’s own fánlì):
- Sān jiā wénzì gǔyì (Three Schools’ textual variants and ancient/odd characters) — collected as a single juǎn at the front.
- The 305-piān-sequence Sān jiā recovery — main body of the work.
- Yì shī (lost poems): poems excised by Confucius, or post-Confucius poems found in classics that the Sān jiā cited (e.g., Yǔ wú zhèng, Gǔ zhōng, Bān).
The work draws on the Hán shū yìwén zhì listing of Sān jiā Shī sources, Hàn shū and Hòu Hàn shū biographies, Lǐ jì, Zuǒ zhuàn, Guó yǔ, the philosopher-classics, Shǐ jì, Bái hǔ tōng, Shuō yuán and Xīn xù, Lǚ shì chūnqiū, Liènǚ zhuàn, the xiǎo xué tradition (Liú Xī’s Shì míng, the Shuō wén), Shū Lǔ shī shí jīng (the late-Hàn Cài Yōng Shī stele), and so on — anchoring source-attribution where possible. The Sìkù tíyào registers approval of Fàn’s zēngrù (expansion) — an additional 60–70% beyond Wáng — and of the yī sānbǎi piān wéi gāng (305-piān-as-frame) reorganization.
The Sìkù editors, however, register two structural objections:
- The work’s title is Sān jiā Shī shí yí, but the inclusion of the gǔwén kǎo yì (ancient-text textual-variants) is anomalous, since these variants are not all from the Sān jiā; conversely, the yì shī (lost poems) are not specifically Sān jiā-attributed, and yet are also included — this is míngshí xiāngguāi (name and content not matching).
- A specific error: Zhāng Chāo’s 張超 Qiào qīngyī fù 誚青衣賦 — which only contains the line “Guān jū = Bìgōng (Duke Bì) zuò” as part of the broader fù — Fàn cites it as if Zhāng Chāo had a separate Shī-explanatory work; in fact this is a single passing line within the fù. (The Sìkù editors’ note: the fù-citation should be quoted.)
The Sìkù conclusion: Fàn’s work is more comprehensive than Wáng Yìnglín’s. The work was followed by parallel projects: Yán Yúdūn’s Sān jiā yí shuō in the appendices of his Dú Shī zhì yí (KR1c0059); Huì Dòng’s Jiǔ jīng gǔ yì 九經古義 and Yú Xiāokè’s Gǔ jīng jiě gōu chén 古經解鈎沉 — but Fàn’s is judged the most comprehensive.
Tiyao
Your servants etc. respectfully present: Sān jiā Shī shí yí in 10 juǎn. By the guócháo (Qīng) Fàn Jiāxiāng. Jiāxiāng’s zì Héngzhōu, native of Kuàijī. Qiánlóng jiǎxū (1754) jìnshì, served as Liǔzhōu fǔ zhīfǔ. In Hàn times, the schools transmitting the Shī were four. The Suí shū jīngjí zhì says the Qí Shī was lost in Wèi, the Lǔ Shī in Xī Jìn — only the Hán Shī survived. The Sòng Tàipíng yùlǎn often cites the Hán Shī; the Chóngwén zǒngmù also catalogues it. Liú Ānshì and Cháo Yuèzhī still occasionally cite its lost readings — but by the Nándù (Southern Crossing of the dynasty) the Confucians no longer mention it; we know it was lost in the Zhènghé / Jiànyán transition.
Since Zhèng Qiáo on, men-and-men labored to put forth new readings, taking attack on the Hàn Confucians as ability; even Máo zhuàn and Zhèng jiān tottered and could barely stand on their own — only by the Tàixué bǎn (Imperial Academy edition) cutting were they fortunately preserved. The Sān jiā lost-text was thereby scattered and could not be recovered. Wáng Yìnglín, at the end of Xiánchún, began to gather the surviving traces and edit a Shī kǎo in 3 juǎn — yet the pioneering work was difficult to perfect, with many gaps. He also added the yì shī piānmù, miscellaneously gathering pseudepigraphic readings of various philosophers without much selection.
Jiāxiāng’s work, drawing on Wáng’s book with re-editing, expands by six-seven-tenths. Taking the 300-piān as frame and the Sān jiā’s lost readings paired per piān, it is also more easily surveyed than Wáng’s recension which had the three schools each as separate piān. Only that, named Sān jiā Shī shí yí, the gǔwén kǎo yì — not all Sān jiā texts — should at least have been appended; and the yì shī not tied to Sān jiā should also have been excised; collected together, it inevitably loses to greed-for-quantity. Further: at the piān-front, míngshí xiāngguāi (name and content not matching) is also a violation of tǐlì. As for Zhāng Chāo’s “Guān jū wéi Bìgōng zuò” entry, the language originally comes from his own Qiào qīngyī fù, not from a separate text by Zhāng Chāo on jīng explanation; based on the Shī bǔ zhuàn’s recording “Zhāng Chāo says…”, it is recorded without quoting the fù text; collection of evidence is occasionally not exhaustive.
But compared to Wáng’s book, his abundance is much greater. Recently Yán Yúdūn’s Shī jīng zhì yí contains a Sān jiā yí shuō one-piān; further, Huì Dòng’s Jiǔ jīng gǔ yì and Yú Xiāokè’s Gǔ jīng jiě gōu chén on Sān jiā all have their excerpts; for comprehensiveness none is up to this work. Qiánlóng 44 (1779), 4th month, respectfully collated. Chief Compilers: Jì Yún, Lù Xīxióng, Sūn Shìyì. Chief Editor: Lù Fèichí.
Abstract
The Sān jiā Shī shí yí is the principal mid-Qiánlóng-era systematic expansion of Wáng Yìnglín’s pioneering Sòng-period Sān jiā Shī recovery work (Shī kǎo). Composition is precisely datable: completed in Qiánlóng 25 (1760) per Fàn’s self-preface dated cháng zhì + 10 days. The work expanded Wáng’s coverage by 60–70% and reorganized the structure into a Máo Shī-piān-sequence frame with the three schools’ fragments bundled by piān. The work was widely respected and influential in late-Qiánlóng Shī studies; it formed the basis for the more thorough Sān jiā Shī recovery work of Chén Qiáozōng 陳喬樅 and others in the late Qīng.
Methodologically the work is at the high-tide of the high-Qing kǎozhèng tradition’s recovery of the Hàn-period classical literature: an attempt to dislodge Máo (which had become the unique-canonical Shī tradition by the post-Sòng period) from its near-monopoly position, and to restore the textual pluralism of the Hàn-period Shī world. Fàn’s complementary work Shī chén (KR1c0063) is the verse-by-verse commentary that puts these Sān jiā recoveries to interpretive use.
The Sìkù editors’ two structural objections — the inclusion of gǔwén kǎo yì and yì shī under a title that promises only Sān jiā recovery, and the misattribution of Zhāng Chāo’s fù-line — register the high standards of evidentiary discipline that high-Qiánlóng kǎozhèng had reached by 1779: even a substantial and influential work was held to a strict standard of míngshí xiāngfú (name and content matching).
Translations and research
No translation. The work is treated centrally in: Lín Qìngzhāng 林慶彰 et al., Sān jiā Shī yánjiū (modern monograph). For broader treatment of the Sān jiā Shī recovery tradition, see Anne Behnke Kinney’s chapters on Hàn-dynasty Shī in Cambridge History of Chinese Literature; David Knechtges’s various articles on Hàn-period scholarship.
Other points of interest
The work’s fánlì (Editorial Principles) — included in the Sìkù-recension _000.txt — gives an unusually detailed methodological statement: nine numbered principles covering source-attribution, character-variant handling, the yì shī category, the four-school transmission history, and so on. The Sān jiā Shī yuánliú (transmission histories) section that precedes the fánlì is itself a substantial reference resource on the four Hàn-period Shī schools and their pupil-lineages — quoting Hàn shū, Lù Démíng, Hàn shū yìwén zhì, and so on. This methodological apparatus is comparable in scope to the fù lù of Yán Yúdūn (KR1c0059).