>The Spoofed Signatures

Run the integrity checker on the war canon and nearly every commit signature comes back spoofed — including the ones generals had studied as scripture for a thousand years.

Chinese military literature has a forgery problem. When Ji Yun's team reached juan 99–100 — filed under Masters (子部), subcategory Military Strategists (兵家類) — they faced a tradition in which nearly every major text claimed to be older than it was, signed by a more famous author than had written it, and padded with material that didn't belong.

Of 67 texts reviewed, 20 made the main catalog and 47 went to the cunmu (存目). The preamble opens with a flat acknowledgment: military writing existed in antiquity, but "everything attributed to Feng Hou and earlier figures is fabricated" (風后以前之書皆偽). Not diplomatic hedging — a senior bibliographer telling you the tradition lied about its own origins. What makes the compilers remarkable isn't spotting forgeries; scholars had questioned attributions for centuries. It's that they applied a systematic verifier and documented every step so later readers could check the reasoning.

The Liu Tao (六韜), the Six Secret Teachings, is attributed to Jiang Taigong (姜太公), the legendary advisor who helped overthrow the Shang around 1046 BCE. If genuine, one of the oldest military texts in existence. It was canonical — Song Emperor Shenzong put it in the Seven Military Classics (武經七書) in 1080, making it exam material for officers. Generations studied it as scripture. The compilers dismantled the signature in layers.

Bibliographic. The Han Shu bibliography lists no "Liu Tao" under military texts; it lists a "Zhou History Six Tao" (周史六弢) under Confucian texts, dated to the Warring States. The received Liu Tao and the Han-era text are not the same book. Vocabulary. The phrase "avoid the main hall" (避正殿) describes a ritual practice that only began in the Warring States — five centuries too late for the early Zhou. Institutional. The rank "general" (將軍) appears throughout; the compilers note it first surfaces in the Zuo Zhuan and didn't exist in the early Zhou military system. Technical. On the Yinfu (陰符) chapter's signal-tally system they deliver their most contemptuous line: 偽撰者不知陰符之義，誤以為符節之符，遂粉飾以為此言，尤為鄙陋 — "the forger did not understand the meaning of yinfu, mistaking it for a physical tally, then embellished it with this nonsense — particularly crude."

And yet they kept it in the main catalog: "those who discuss military affairs have always cited it" (談兵之家恒相稱述). Too entrenched to remove. So they documented the forgery in full and let the canon stand. A bureaucrat's compromise: the record is corrected even if the institution can't be.

The San Lue (三略), the Three Strategies, arrives wrapped in legend: Zhang Liang (張良), architect of the Han founding, supposedly received a mysterious book from an old man on a bridge — later identified as Huangshi Gong (黃石公), the "Old Man of the Yellow Stone." The story is in the Shiji; Sima Qian tells it. The compilers don't dispute the legend — they dispute the text. Their verdict: "not ancient in its phrasing" (文義不古) and "certainly a later fabrication" (當亦後人所依託). The vocabulary belongs to the late Warring States or early Han, not the pre-Qin world the legend places it in. The simplest explanation: compiled in the Han from earlier and contemporary sources, attached to the Zhang Liang legend for prestige.

The Li Weigong Wendui (李衛公問對) — supposedly a dialogue between Tang Taizong and his general Li Jing — is the case where the compilers went furthest. They didn't just prove forgery; they identified the forger. Multiple Song sources name Ruan Yi (阮逸), a scholar who also fabricated a text attributed to the philosopher Wang Tong and an I-Ching commentary attributed to Guan Ziming. Su Shi reported that his father had seen Ruan Yi's drafts. Three independent sources, same man — a serial forger caught by the paper trail.

But the compilers pushed back on dismissing it entirely. The text "distinguishes orthodox and unorthodox formations, maps attack and defense, and occasionally grasps the subtleties of military thought." As Zheng Yuan put it: 問對之書雖偽，然必出於有學識謀略者之手 — "though the text is forged, it must have come from a hand with real learning and strategic sense." A forgery by a competent thinker outranks a forgery by a hack.

Below the famous forgeries sits a stratum the compilers treated with undisguised contempt. The Wo Qi Jing (握奇經) claims an eight-formation array transmitted from the mythological minister Feng Hou; the compilers traced it to Tang enthusiasts who reverse-engineered Zhuge Liang's Stone Sentinel Maze and projected it backward to the Yellow Emperor. The text is exactly 384 characters — the same as the total lines in the 64 hexagrams. Not coincidence: the formation was conceived as a cosmological diagram, eight trigrams mapped to eight tactical positions. The compilers called it numerological theater, not military science.

The cunmu is worse. They describe the late Ming output as "especially squalid and chaotic" (尤為猥雜): wooden men on horseback with firecrackers in their bellies ("virtually a children's game" — 殆於兒戲), proposals to grease ship decks with egg and tung oil so boarders fall, and astrological battle-timing systems that amount to nothing. Each follows the same pattern — attribute it to someone famous, fill it with impressive nonsense, hope no one checks.

The verdicts resolve into a clear toolkit, worth stating because the West wouldn't codify these until the nineteenth century. Vocabulary dating — words have lifespans; a Zhou-era text using Warring States terms isn't Zhou. Citation chain analysis — checking every text against the Han, Sui, Tang, and Song bibliographies; a gap is evidence of fabrication after that date. Internal consistency — no genuine text contains institutions or place-names postdating its claimed composition. Stylistic analysis — distinguishing texts that "read ancient" (文義古) from those that don't, as obvious to a trained reader as Elizabethan English claiming to be Anglo-Saxon. Author identification — when independent sources name the same forger, the identification stands.

Western textual criticism formalized its methods in the late eighteenth and nineteenth centuries — Wolf's Prolegomena ad Homerum (1795), Lachmann's manuscript stemmata. The Siku scholars were doing the same work, at the same time or earlier, on a larger corpus: 10,000+ texts, the military section alone covering 67, each assessed for authorship, dating, integrity, and value. The difference is institutional. Wolf and Lachmann founded academic disciplines; Ji Yun's team produced a catalog filed in the palace library. The methods were parallel; the institutional afterlives diverged.

The compilers' final position is neither purist nor permissive — it's pragmatic. A forged text that teaches real strategy beats a genuine text that teaches nothing. They documented the truth, then let the institution make its own decisions.