>The Query Manual

A 200-volume database is only as useful as your ability to query it. Here is the record schema, the flag vocabulary, and the access endpoints.

Every entry follows the same schema. Title (書名) — usually the full form; variants noted in the body. Author and dynasty (著者, 朝代) — the attributed author and their dynasty; if the attribution is disputed, the entry says so, often with hedges like "attributed to" (舊題) or "supposed to be by" (舊本題), the first signal the compilers have doubts. Juan count (卷數) — the number of fascicles in the reviewed edition; a discrepancy from earlier bibliographies gets flagged and explained. Critical review (提要) — the substantive part: a paragraph to a full page of classical Chinese summarizing contents, assessing quality, evaluating authenticity, noting relationships, and rendering a verdict.

The catalog has two tiers per category. The main catalog (著錄) holds texts selected for full transcription into the library — the compilers' endorsement. The cunmu (存目, "preserved title") holds texts reviewed but not selected. Relegation to the cunmu isn't always condemnation — some entries describe useful-but-derivative or interesting-but-short texts — but many are critical, identifying the text as forged, incompetent, or worthless, and the cunmu is where the sharpest demolitions appear. Always check the tier first: it tells you the bottom-line verdict before you read a word of the review.

The reviews use a specialized evaluative vocabulary. Learn the flags and opaque classical Chinese becomes a readable reference.

Positive: 精審 (jingshen) — "precise and careful," the highest praise for scholarly rigor. 有所發明 (you suo faming) — "has original insights," originality grounded in evidence. 足資考證 (zu zi kaozheng) — "sufficient for evidential research," provides usable data even if otherwise flawed.

Negative: 穿鑿附會 (chuanzao fuhui) — "forced and fabricated," the standard dismissal for interpretations that twist evidence. 鄙俚淺陋 (bili qianlou) — "vulgar, shallow, crude." 無所發明 (wu suo faming) — "no original insight," i.e. derivative. 偽託 (weituo) — "falsely attributed," the standard forgery flag.

Authenticity hedges: 舊題 (jiuti) — "old title says...," reporting an attribution without endorsing it. 不著錄 (bu zhulu) — "not recorded in previous bibliographies," the bibliographic silence argument that treats absence as strong evidence of forgery.

The four-part structure is itself a navigation tool. Knowing where a text is classified tells you what the compilers thought it was for. 經部 (Classics) — juan 1–48: the foundational texts and their commentaries, plus philology; Six Lines draws primarily on the Yijing section (juan 1–10). 史部 (History) — juan 49–85: official histories, chronologies, institutional and geographical records. 子部 (Masters) — juan 86–148: the broadest, covering philosophy, military strategy, law, agriculture, medicine, astronomy, divination, Buddhism, Daoism; Six Lines draws on the military section (juan 99–100), the Legalist section (juan 101), and divination (juan 108–111). 集部 (Collections) — juan 149–200: literary anthologies, collected works, criticism, drama. Each division subdivides into categories (類), and each category opens with a preface (小序) — compressed histories of entire fields, written by scholars who'd just finished reading everything in them.

A practical approach: start with the category preface — it states the compilers' general standards for that field, and individual reviews are written against that framework. Check the tier — main catalog or cunmu. Read the first sentence — it identifies author, period, and basic character; attribution doubts appear here. Look for the evaluative turn — most reviews describe contents first, then shift to judgment, often signaled by 然 (ran, "however"), 惟 (wei, "only/but"), or 其 (qi, "its [flaw is]"); everything after the turn is the verdict. Watch for comparative judgments — "superior to X but inferior to Y," "follows the methods of Z but without Z's insight" — these situate the text in context. Use the cunmu for context — reading what the compilers rejected, and why, clarifies by contrast what they valued.

ctext.org (Chinese Text Project) — the most accessible free source, with transcribed text navigable by juan and category. Wikisource (zh.wikisource.org) — juan-by-juan transcriptions with clean formatting; search 四庫全書總目提要/卷 plus the juan number. CADAL scans (Archive.org) — the China Academic Digital Associative Library at Zhejiang University digitized the original 18th-century manuscripts at 600 dpi. Commercial databases — the Wenyuange Siku Quanshu (文淵閣四庫全書) provides full-text searchable access to both catalog and complete library, via institutional subscriptions.

For researchers: start with prefaces — they provide the framework that makes the reviews intelligible. Use the cunmu — the most candid and entertaining entries, and where the ghost entries (reviews of texts that no longer survive) live. Compare with modern bibliographies — the Siku verdicts are the starting point, not the last word; Wilkinson's Chinese History: A New Manual provides updated assessments correcting the compilers' biases. Read the reviews in Chinese — the classical prose is clear and economical, the vocabulary consistent, the structure predictable; the original carries nuances translation can't. Cross-reference within the catalog — when a review says "see the entry on [other text]," follow it; the cross-references hold the most interesting comparative judgments.

The Siku Quanshu Zongmu Tiyao is a read-only database of ten thousand verdicts, queryable by anyone who learns its schema. Know the record structure, learn the flag vocabulary, and 200 volumes of classical Chinese literary criticism resolve into the most comprehensive reference the tradition ever produced.