A king sits with his chief officer and tells him to be careful. Not careful with money, or with the harvest, or with the border — careful with men. Before you put a man in a job, the king says, you have to know what he actually is, and most of what a man shows you on the surface is something he built to be seen. So watch how he behaves when he is rich. Watch how he behaves when he is poor. Get him drunk and see if his manners hold. Dangle money in front of him and see if his eyes change. Push him away and see if he sulks. Pull him close and see if he gets familiar. The man cannot keep it all hidden forever; if you set up the right situations, the truth leaks out.
That scene — a founder of the Chou dynasty lecturing his officer on how to evaluate, select, and place men — is what "The Offices of King Wen" preserves.1 The text is also called the Wen-wang Kuan-jen, and it survives in two ancient collections, the Yi Chou-shu and the Ta-Tai Li-chi, both probably reaching final form in the third century B.C.2 Sawyer treats it as the high-water mark of pre-modern Chinese personnel evaluation. He says flatly that before Liu Shao's Jen-wu Chih — a work centuries later — "there is nothing to compare with it."3 For anyone whose business is selecting agents, vetting officials, or detecting the man wearing a mask, this is the foundational codex. It is the closest thing the early Chinese tradition has to a complete handbook for the question every intelligence director eventually faces: who is this person really, and where do I put him?
The text is a structured taxonomy for reading and placing men. Its architecture is announced up front: there are seven categories of ability, each category has nine employments, and each employment has six indications.4 The six indications are the diagnostic instruments — the six ways you look at a man. King Wen lists them as: observe his sincerity; test his intentions; look within; observe his appearance; observe his hidden aspects; estimate his Virtue.5
Underneath the structure sits a single working premise, and it is worth being precise about it because the whole method depends on it. King Wen believes character can be fathomed — that "emotions and true personality cannot be concealed."6 When happiness, anger, desire, fear, or sorrow accumulate inside a man, the signs come out whether he wants them to or not.7 But — and this is what makes the text sophisticated rather than naive — the king does not stop at people are transparent. He spends most of his words on the men who actively work to defeat transparency: the ones who build facades, who manipulate their own image, who hide their real nature inside a costume of virtue.8 The text accepts that the truth will leak, and then catalogues, in granular detail, every disguise men use to plug the leak.
So it is two things at once. It is an optimistic theory — the heart shows on the face. And it is a paranoid field manual — here are the forty masks, learn to see through each one.
The engine that drives the whole system is coherence. King Wen's test for a fake is not any single tell. It is the gap between parts. "When words and actions are not of the same category, end and beginning contrary to each other, yin and yang conquer in turn, and the exterior and interior do not cohere... it is said not to be true substance."9 A real man matches himself across contexts and across time. A constructed man, sooner or later, contradicts himself — the loyalty he performs in front of others does not match the way he treats them when no one is watching; the words today do not match the words from years ago; the visible effort does not match the invisible motive.
That is why the method moves a man through situations rather than reading him in a single frozen pose. The first instrument — "observe his sincerity" — works by varying his circumstances and watching whether the same person survives the change. If he is wealthy and noble, watch whether he keeps the rites or grows arrogant. If he is poor, watch whether he keeps his integrity or is driven to crooked acts. If the ruler heaps favor on him, watch whether he stays unspoiled. If he is unknown and stuck, watch whether anxiety eats him.10 Sawyer's own gloss is sharp here: the criteria assume less than ideal human tendencies — the rich usually do get arrogant, the favored usually do exploit their position — so holding proper behavior against those default pulls is what reveals real character.11
Then the method gets active. It does not wait for situations to arrive; it manufactures them. "Show him hardship in order to observe his courage. Immerse him in profit in order to observe whether he isn't greedy. Immerse him in pleasure to observe whether he doesn't become dissolute... Make him drunk in order to observe if he doesn't violate propriety. Keep him at a distance in order to see if he isn't disloyal. Keep him nearby in order to see if he doesn't become too familiar."12 These are deliberate stimuli. You build a pressure and read the reaction against an expected standard. Sawyer flags the heaviest probes as the temptations of wealth and beauty — man's "powerful, often overwhelming desires" — alongside engineered rejection and favor, and even planted false clues or suggestions to see what a man does with bad information.13 This is no longer passive physiognomy. It is interrogation by environment.
The deepest of the six instruments — "look within" — adds a sensory channel most evaluators ignore: the voice. The premise is that inner ch'i shapes sound. "If the mental ch'i is flowery and expansive, then the sound is flowing and dispersed... If the mental state is vulgar and perverse, then the sound is hoarse and ugly."14 You listen to the sound, determine the ch'i behind it, and from the small thing (the timbre) you estimate the large thing (the man). "Estimate the future from the past, the invisible by what is manifest, and the large by the small."15
The richest part of the text — and the part that makes it a spycraft document rather than a piece of court etiquette — is the section "Observe What Is Concealed." Here King Wen states that people have yin and yang aspects, and that many men "conceal their emotions" and "adorn themselves with artifice in order to seek fame."16 He then itemizes the types of hiding, sorted by which virtue the man hides inside:
This is the spine of the whole document for intelligence purposes. It is a catalogue of the good-seeming bad man — the operator who weaponizes a virtue as camouflage. Each entry names the disguise and then names the tell that breaks it. The king's closing instruction ties the method into a single rule: "Observe their yang [visible aspects] in order to examine into their yin [hidden natures]. Investigate them internally in order to appraise them externally."23 Read the visible to reach the hidden. That is the whole craft compressed to a sentence.
Reading a man is only half the system. The other half is placement. The Nine Employments is a role-matching grid: each profile of character is mapped to the office it should hold.24
"In accord with their talents you should employ them," the king says — this is what "employing ability" means.26 The grid is the answer to the second half of the intelligence director's problem. You don't just detect a man's nature; you find the slot his nature fits. The man who is firm and self-reliant in decisions goes to the border, where there is no time to ask for orders. The man who is sensitive and perceptive goes to the guests, where the work is reading a room. Above the nine sit the seven categories — the broadest sorting of who belongs where: for the state use men of noble rank, for villages the true, for offices leaders, for students teachers, and so on.27 Sawyer notes the seven are thin compared to the six indications — titles only, with little elaboration — and reads that imbalance as telling: the energy of the whole text falls on evaluating personality and talent, not on the administrative scaffolding.28
This page is the keystone of the vault's "knowing men" cluster. Other pages describe the problem of reading character or the recruitment of agents in the abstract; this one supplies the actual instrument panel — six channels, a concealment taxonomy, and a placement grid. When a later page says an agent must be vetted, this is the document that says how.
It also gives the vault a clean distinction it otherwise lacks: passive observation versus active probing. Half of King Wen's method is watching a man across the situations life hands him. The other half is building the situations — the drunk test, the profit test, the planted false clue. That split is the seed of the whole espionage-tradecraft idea, sitting fully formed in a third-century-B.C. court lecture.
Take one probe from the "actively test" list and follow it all the way down: "Make him drunk in order to observe if he doesn't violate propriety."29
On the surface it reads like folk wisdom — in vino veritas, the drunk tells the truth. But notice what the king has actually constructed. He has a working theory that the five emotions accumulate inside and leak out as visible signs regardless of intent.30 He has a complementary problem: skilled men build facades precisely to suppress those leaks.31 Drunkenness is the solvent. It does not create the man's true nature; it dissolves the facade that was holding the nature down. The greedy man who has trained himself to perform restraint loses the muscle that performs it. The man hiding fear inside a costume of "purity and courage" — "within they are afraid, without they are braggarts" — cannot keep the brag aligned with the fear once the controlling mind slackens.32
And the king is not reading drunk behavior naively either. The verdict is not "he got rowdy, therefore bad." The verdict is coherence: does the drunk man violate propriety — does the sober facade and the drunk reality fail to cohere? A man whose drunk self matches his sober self has passed. A man whose mask slides off the moment the ch'i is clouded was wearing a mask. This is exactly the test Sawyer flags as the section's logic: King Wen "provided methods for actively setting up stimuli to elicit reactions that might then be judged for internal coherence, as well as appraised against external standards."33 One throwaway line about a banquet turns out to encode the entire epistemology: leak theory plus facade theory plus the coherence verdict, all in seven words.
You are the T'ai-shih, and the king has just told you to be cautious. There is a man you are considering for the treasury. On paper he is perfect: incorruptible reputation, sober manner, talks often about his honesty.
You do not trust the paper. You start watching across contexts. In mourning, is his grief real or performed? Among his friends, is he faithful or transactional? You note that he tells you, unprompted, how scrupulous he is — and a small alarm goes off, because the men who hide in purity are exactly the ones whose "words bespeak integrity so that they may be taken as their character."34
So you build a test. You leave him alone with an accounting where a sum could quietly go missing and no one would know — a planted opportunity, a false clue that the money is untraceable.35 You "immerse him in profit" and watch.36 You do not watch only whether he steals; you watch the texture of him while the temptation sits there — the voice, whether it stays "correct and placid" or goes hoarse, because the ch'i of trustworthiness has its own sound.37 Days later you raise the matter casually, "revisit his minor words," and check whether today's account squares with the one he gave before.38
If everything coheres — sober and tempted, public and private, this week and last — you have found a man of "expansive knowledge" or one "careful in investigations, incorrupt and pure," and the Nine Employments tells you exactly where he goes: the treasury, dividing wealth and making rewards.39 You have not just judged him. You have placed him.
You know the method is failing when you find yourself reading a man in a single pose. One impressive interview, one banquet, one letter of recommendation — and you decide. King Wen's whole architecture is built against the single snapshot. If you are not moving the man through varied situations and varied time, you are not running the method; you are admiring a portrait the man painted of himself.
You know it is failing when the virtues come announced. The real article does not narrate itself. "The circumspect and good... speak little but put it into action... They have knowledge but do not flaunt it."40 The man telling you how loyal, how filial, how pure he is, is showing you the precise location of the mask — these are the men who "flaunt their filial reputations in order to profit."41
You know it is failing when you mistake consistency of performance for coherence of substance. A skilled faker is internally consistent within the role he is playing — that is what makes him skilled. The tell is not whether the performance is smooth; it is whether the performance survives a probe it did not see coming. "Oppose them with sudden affairs" — the pure and decisive will not be afraid; the vulgar and artificial will be "easily scared," their changes "easily known."42 If you never spring the sudden affair, you never learn which one you have.
And you know the whole apparatus has failed at the back end when you read a man correctly and then place him wrong — when you put the border-temperament man in the guest hall, or the room-reader on the frontier. The Nine Employments exists because accurate diagnosis followed by bad assignment wastes the diagnosis entirely.
The text contains a real internal tension that Sawyer does not smooth over, and neither should we. King Wen opens by asserting transparency — emotions cannot be concealed, the true personality always shows.43 Then he spends the bulk of the lecture on men who successfully conceal long enough to forge reputations and capture office.44 If concealment were truly impossible, the catalogue of concealers would be unnecessary. The text lives inside that contradiction rather than resolving it: transparency is the long-run law (leak theory), but in the short run a clever man can dam the leak well enough to deceive a careless evaluator — which is exactly why the active probes exist. The probes are the king's tacit admission that passive transparency is not enough.
There is also a provenance question. Sawyer notes the two surviving versions disagree on basics — including whether it is King Wen instructing the officer or, less plausibly, the Duke of Chou instructing King Wen — and that the translation leans on the more expansive Ta-Tai Li-chi version, emended against the Yi Chou-shu and reconstructed with heavy recourse to modern scholarly editions because the characterizing adjectives are "often puzzling."45 The attribution to King Wen himself is almost certainly pious framing on a Warring States compilation, not a transcript of a Chou founder. The ideas draw on "the common knowledge of the Warring States period about human tendencies," measured against an idealized figure — which is the real dating, whatever the cover story says.46
Open question: the seven categories are conspicuously thin. Sawyer offers three possibilities — material was lost, the king never expanded them, or the emphasis simply and naturally fell on evaluating personality and talent rather than on titles.47 We cannot know which. What we can say is that the surviving balance — six indications rich and detailed, seven categories skeletal — is itself the clearest evidence of what this text was for: it was built to read men, not to organize a bureaucracy.
Sawyer's framing pulls in two directions at once, and the friction is productive. As a translator he is scrupulous — he tells you the versions disagree, that he emended, that the adjectives are puzzling, that he leaned on Kao Ming and Li Hsüeh-ch'in to make sense of the character types.48 But as a synthesizer arguing for a continuous Chinese intelligence tradition, he wants this text to stand as a "summation and expression of the general approach to evaluating," a keystone he can connect forward to Liu Shao and backward to the broader "knowing men" project.49 Those two impulses do not always sit easily: the careful philologist admits the text is a reconstructed composite of uncertain date, while the tradition-builder treats it as a stable monument with "nothing to compare with it."
The honest reading holds both. The Wen-wang Kuan-jen is genuinely the most complete pre-Liu-Shao evaluation codex we have — Sawyer is right about that. But its monumentality is partly an artifact of survival and partly of Sawyer's own organizing frame. Where his cross-dynasty continuity claims need the most independent verification is precisely the claim of unbroken transmission — that later evaluators read and built on this text rather than independently re-deriving similar folk-psychology from the same Warring States common stock. The convergence of methods across centuries is real; whether it is descent or parallel reinvention is exactly the kind of continuity claim a translator-synthesizer is inclined to read as the former.
To Physiognomy and Character Reading. Both pages chase the same prize — reading the inner man from the outer surface — but they reach for different surfaces, and the difference is the insight. Physiognomy reads the fixed surface: the bones of the face, the standing form, the features a man was born with and cannot change. King Wen's six indications read the dynamic surface: behavior under varied and engineered pressure, voice under emotional load, the gap between this week's words and last week's. The structural tension is about agency. Physiognomy assumes the body honestly betrays the fate stamped into it. King Wen assumes the man is an active adversary who builds facades on purpose, so the evaluator must build counter-situations to break them. Put the two together and you get something neither produces alone: a layered read where the static features set a prior and the dynamic probes test whether the man is living up to, or actively masking, that prior. Physiognomy says what he probably is; the probes say what he does when he thinks he can get away with something. The faker can paint his face calm; he has a much harder time keeping the painted calm coherent across a drunk banquet and a planted temptation. The body is the hypothesis; the probe is the experiment.
To The Semblances Problem. The semblances problem is the general trap that surfaces resemble each other — that a thing and its counterfeit present the same face, so appearances mislead. King Wen's concealment taxonomy is the applied version of that trap, scoped to people and weaponized by a conscious agent. The man who "hides in benevolence" is a living semblance: he presents the exact face of the virtuous man, and the resemblance is deliberate, manufactured to capture office and fame. What King Wen adds to the abstract semblances problem is a resolution procedure. Semblances mislead in a single frozen view; coherence-across-pressure is the method that separates the thing from its counterfeit, because the counterfeit was built to survive one look and cannot survive a sequence of unexpected ones. The insight that neither page reaches alone: the semblances problem is unsolvable in a snapshot and partly solvable in time. You cannot tell the real Worthy from the performed Worthy by looking harder at one moment. You can tell them apart by watching whether the performance holds when you change the lighting — wealth, then poverty; sober, then drunk; favor, then rejection; a true account, then a planted false one. Time and engineered variation are the only solvents the semblances problem has.
Sharpest implication: The whole document is built on a wager that facades are cheap to build and expensive to maintain. A man can paint any single virtue onto his face for one occasion. What he cannot easily do is keep forty separate paintings coherent across drunk and sober, rich and poor, public and private, this year and last. King Wen's method does not try to see through the mask directly — it makes the mask do more work than a mask can sustain, and waits for a seam to split. That is a transferable principle far beyond the Chou court: you defeat a sophisticated performance not by scrutinizing it harder but by multiplying the contexts it has to survive.
Generative questions: