Can the Gods Be Deceived?

Part OneWhat the Book Says About Itself

In 1424 a Ming official sat down to write a preface for a book about two gods.

The gods were brothers. Six centuries earlier they had been men — 徐知證 Xú Zhīzhèng and 徐知諤 Xú Zhī'è, sons of a Wu dynasty chancellor, soldier-princes who marched into Min at the head of an army in the 940s and were, in the way of such things, buried by grateful villagers and then slowly promoted. A shrine. A rescript. A title. Another title. By the time the preface was written they were 洪恩靈濟真君 Hóng'ēn Líng Jì Zhēnjūn, Perfected Lords of Vast Grace and Numinous Deliverance, and they had a temple in Beijing.

They also had a literature. Fourteen chapters of it, compiled in 1305 by 陳夢根 Chén Mènggēn, gathered under the title 徐仙翰藻 Xú Xiān Hànzǎo — Hànzǎo, "brush-flowerings," the writings of the Xu Immortals. Two hundred–odd pieces: stele inscriptions, prefaces, hymns, rhapsodies, subscription appeals, proclamations against demons.¹

The preface-writer describes this book in a single clause that ought to stop a reader cold:

皆錄神之筆也。

Jiē lù shén zhī bǐ yě.

All of it records the god's brush.²

Not written about the gods. Written by them. Every genre in the collection — 碑銘 bēimíng stele-inscriptions, 傳 zhuàn biographies, 跋 bá colophons, 序 xù prefaces, 啟 qǐ letters, 頌 sòng hymns, 疏 shū subscription appeals, down to the Turtle Peak Rhapsody 鰲峰賦 and the encomia painted on the walls 壁像贊 — the preface files them all under one origin. The brothers wrote them. Personally. After death.

The mechanism by which a dead man writes is not a mystery in this tradition, and the preface does not treat it as one. It is called 扶乩 fú jī, supporting the planchette — also 扶鸞 fú luán, supporting the phoenix, or 降筆 jiàng bǐ, the descent of the brush. A forked stylus 乩架 jī jià of peach or willow, held by two people over a tray of fine sand 沙盤 shā pán; the deity descends into one of them and moves the stylus; the characters scratched in the sand are read aloud by a caller 唱生 chàng shēng and copied down by a recorder 記錄 jì lù. The apparatus at full strength ran to six, the 六部 liù bù. What the recorder produced was a text, and the text could then be edited, anthologized, and printed.

So the Hànzǎo is, on its own account, a transcript.

Part TwoA Theory of Leakage

What makes the 1424 preface remarkable is not the claim. Spirit-writing collections are not rare, and prefaces overclaim as a matter of course. What is remarkable is that the preface stops to explain why gods write at all — and the explanation is not one a modern reader would have guessed.

惟神得兆朕之先，暫否其德，則泄為文章，使虛空縹緲之筆，燎然而駭俗，粲然而耀衆。

Wéi shén dé zhào zhèn zhī xiān, zàn pǐ qí dé, zé xiè wéi wénzhāng, shǐ xūkōng piāomiǎo zhī bǐ, liáo rán ér hài sú, càn rán ér yào zhòng.

Only the god, having priority over omens and portents, when his virtue was for a time blocked, then leaked out as literary composition — causing the brush of the empty void and the misty-indistinct to blaze up and startle the vulgar, to glitter and dazzle the multitude.

泄 xiè is a hydraulic verb. Seepage, discharge, the escape of something under pressure. 否 pǐ is the Yijing hexagram of obstruction, standstill, the condition in which heaven and earth do not communicate.

The preface is proposing that divine efficacy behaves like water in a closed system. When the god's proper channel — the giving of rain, the turning back of plague, the ordinary business of a territorial guardian — is stopped, the force does not dissipate. It finds another outlet. It comes out as writing.

This is a claim about the relationship between failed thaumaturgy and literary production, and it is stated by the tradition about itself, without embarrassment. The god writes because he is, for the moment, prevented from acting. Text is what efficacy becomes when efficacy is dammed.

And the instrument through which it emerges gets a name that is not mechanical at all. Not the stylus, not the sand-tray, not the medium's wrist. 虛空縹緲之筆 xūkōng piāomiǎo zhī bǐ — the brush of emptiness and mist. A writing implement made of the same stuff as the void.

Part ThreeThe Problem This Creates

A tradition that claims its entire literature was dictated by a god has a difficulty, and the difficulty is not the one an outsider expects.

The outsider's objection is: this is a man moving a stick through sand. That objection was available at the time and was made. Six centuries after the preface, the Qing scholar 紀昀 Jì Yún would put it with lethal economy — spirit-writing, he observed, tracks the skill of whoever is holding the stylus. Get a good calligrapher and the calligraphy is fine. Get a good poet and the poetry is fine. Get a man who can do neither, and the piece comes out haltingly.³ The tray writes at the level of the hands.

But that objection, however sharp, was never the tradition's operational problem. The operational problem was competition.

If a god can descend into a sand-tray at Áo Fēng, he can descend into a sand-tray anywhere. Every shrine with a forked stick and a literate scribe could produce a text and file it under divine authorship. The technology is cheap, portable, and unlicensed. In the Song it spread widely enough that both 沈括 Shěn Kuò and 蘇軾 Sū Shì wrote about encountering it, and by the thirteenth century a great many gods were writing a great deal.

Which meant that the question is this text divinely authored? had, in practice, collapsed into a different question: whose planchette?

And whose planchette is not, at bottom, a question about paperwork. It is a question about authority over the real — about who holds the standing to certify that a given text is what it claims to be. The licensing regime is the state's administrative answer to that question, but the question is older and deeper than any plaque: when a claim cannot be checked against the thing it reports — when the referent is a god, who cannot be produced — the only remaining question is who is qualified to be believed. The plaque and the title are how an empire files that question. They are not how it is settled.

That is a question about credentials. And credentials, in the Chinese religious economy of the period, were not a private matter. A cult with a bestowed plaque 賜額 cì é and a conferred title 封號 fēng hào was 正祀 zhèngsì, legitimate sacrifice. A cult without them was 淫祀 yínsì — excessive, illicit, and liable to be shut down.⁴ The Xu brothers' documented career reads, from a certain angle, as nothing but a licensing history: shrine in 983, incense divided in 1117, imperial title in 1238, elevation in 1485. Each step is a piece of paper.

The Hànzǎo is the file that supports the application.

Part FourTwo Texts in the Third Chapter

Which brings us to the point of this essay.

In 卷三 juǎn 3 of the Hànzǎo, among the rhapsodies, there sits a cluster of four prose and verse pieces that do something the rest of the collection does not. They argue.

- 自辨文 Zì Biàn Wén — Text of Self-Defense - 塞謗文 Sāi Bàng Wén — Text to Stop Slander - 非神賦 Fēi Shén Fù — Rhapsody on Not Being a God - 莫毀碑辭 Mò Huǐ Bēi Cí — Words Against Destroying the Stele

Four documents, in one chapter, in which the deity defends his own legitimacy — and does so, necessarily, through the very medium whose legitimacy is at issue. The planchette arguing for the planchette. The god who is accused of not existing, replying.

This essay reads two of them.

塞謗文 is a prosecution. It opens as prose, sets out what a genuine descent looks like and what a fraudulent one looks like, and then — at the exact moment it moves from argument to enforcement — announces 姑作筆 gū zuò bǐ, "for now I make the brush write," and shifts into four-character verse. Buried in its prose is something easy to read past: a working rubric.⁵ Seven tests a legitimate planchette must pass, and six marks by which a fraudulent one may be known. It is, in effect, a professional standards document, issued by the standard-setter, in the standard-setter's own hand.

非神賦 is a defense. Its title names the charge: 非神 fēi shén, "not a god." It answers in 賦 fù — the rhapsody, the 楚辭 Chǔ cí-descended form of first-person lament and self-vindication that goes back to 屈原 Qū Yuán — and it answers, in the end, by declining to answer: 予嘿嘿而何辭 yú mò mò ér hé cí, "I, in silence — and what is there to say?"

One text sets the standard. The other meets the challenge. Between them they constitute something I have not seen described: a revealed corpus arguing, in revealed texts, for the validity of revelation.

The question the essay pursues is the one those two documents put on the table and cannot step outside of.

Can a revealed text establish its own revelation?

The apparatus knows the question is circular. It answers anyway — and the form of the answer is itself the argument, because the reply arrives in the rhapsodic register that this tradition reserves for the moment when the god's own hand is on the brush.

Part FiveThe Rubric in the Composition to Stop Slander

The Sāi Bàng Wén 塞謗文⁶ is the third chapter's standards document. It is short — eight prose sentences and a twelve-couplet verse — and it does one thing: it tells you how to distinguish a real planchette from a false one. What it never does is tell you how to distinguish a true message from a false one. That omission is not an oversight. It is the whole design, and recovering why turns the text from a curiosity into evidence about how a revealed cult defended its franchise.

Read as a list, the criteria look scattered — a remark about the god's habits, a note on the petitioner, a line on the equipment, a paragraph on the frauds. That is how they have been counted here so far: seven tests a true descent must pass, six marks by which a false one is known. But the seven are not seven independent hurdles. Read as a schema, they resolve into a single structure: a qualifications regime organized by role. Every party to a consultation is assigned a test, and the tests run down the transaction in order.

The four stations

The god goes first, and qualifies himself by station. "Since I received my office among the immortals," the text opens, 予自領仙職以來 — and the verb is 領, the ordinary word for taking up an appointment. Divinity is introduced not as a nature but as a post, held on the same terms a prefect holds a prefecture. The god then dates his own conduct to the temple's calendar: after the 普度, the universal-deliverance rite the cult administered, he has "not for a single day kept company with men." This is usually read as ascetic withdrawal. It is not. A 普度 is the most public event a cult stages — communal, funded, attended by the whole district — and the god is measuring his restraint against it. I do not meet people in rooms. The first qualification is a matter of visible comportment, and it sets the pattern for everything after: the god's standing rests on his being seen to operate in the open.

The petitioner qualifies by grade. "Unless it is the man himself, undivided, who sets up the resonance, no portent comes back" — 苟非其人精純所響，徵兆莫報. The two characters rendered "undivided," 精純, are refinement words: unmixed grain, silk of a single color. They belong to the vocabulary of quality, not of purity — the text nowhere reaches for the language of ritual cleanliness. What the petitioner must bring is not a clean soul but a concentrated one, and the reason is mechanical. The consultation works by resonance, 響, and resonance operates by category: like answers like. A request of mixed or low grade sets up no clear resonance, and nothing comes back. Note where this places the burden. When the apparatus is silent, the fault is assigned in advance to the petitioner's grade. The system cannot fail; only the applicant can.

The procedure qualifies by responsiveness. "Utterly impartial, without private interest: put a doubt and there is an answer; the cup replies as an echo replies" — 至公無私 … 如響斯答. 至公無私, "utterly impartial, without private interest," is bureaucratic virtue, the phrase one uses of a good magistrate; the god's disinterest is his office-holding virtue, not his sanctity. And the echo does real work here. An echo returns what was actually sent, not what the sender claims to have sent. A procedure built on echo is therefore one that cannot be talked into the wrong answer — which is exactly the property the next station will accuse the frauds of lacking.

The operators qualify by conduct, and this is where the text spends its polemic. The frauds are named, in one dense sentence, by six marks — and every one of the six is procedural. They work unwitnessed: "peer in at their door and the room is empty of anyone," 闚其戶，閴其無人. They dress up their pronouncements, 文飾其說. They contrive and angle for praise, 釣誇. Not one of the six marks alleges that what the frauds say is false. The charge is never that the message is wrong. It is that the process was improper — unwitnessed, ornamented, self-serving. The text is an inspection report, and it inspects the procedure, never the product.

The one test no one can run

Three of the four stations are inspectable. Anyone can check whether a session had witnesses, whether the claims were ornamented, whether the operator promoted himself, whether the apparatus answered. These are public facts, verifiable by a qualified observer — which is the point of the pearl the text reaches for at its center: "the true and the counterfeit can be seen," 真贋可見. The discrimination is ocular. You look, and you can tell.

One station is not like this. The petitioner's grade, 精純, cannot be inspected by anyone. There is no external test for whether a request was "undivided." And that single uninspectable term is precisely the one the text loads every failure onto. When nothing comes back, it was not that the god was absent or the apparatus false — it was that the petitioner was not of the grade. A rubric that is otherwise wholly procedural, wholly about checkable form, keeps exactly one unfalsifiable criterion in reserve, and points it at the customer. This is not a flaw in the design. It is the pressure-relief valve that lets a procedural system absorb the one thing procedure cannot guarantee: results.

Why the message is never tested

Step back and the omission at the center of the text becomes legible. The Sāi Bàng Wén cannot test the truth of a revelation, because there is no test for it. A message from a god is unverifiable by construction; that is what makes it a message from a god. So the apparatus does the only thing available to a standards-setter who cannot inspect the product: it inspects the producer. It cannot certify the message, so it certifies the procedure and the operator's standing to run it.

This is the ordinary logic of expert authority everywhere. We cannot evaluate a bridge before we cross it, so we license the engineer. We cannot audit a diagnosis from the waiting room, so we credential the physician. The proof of qualification stands in for the proof of correctness, because the proof of correctness is unavailable. The Sāi Bàng Wén is doing exactly this, and the only remarkable thing is the domain: the product being warranted is contact with the dead, and the credential is command of the Confucian canon.

Which is why the text is built the way it is. It does not merely describe the operator's qualification; it performs it. In twenty short units the god quotes or invokes the Analects twice, the Guoyu, the Book of Changes, the Odes twice, Mencius, the Records of the Grand Historian, the History of the Later Han, the Zuo Commentary, and the Huainanzi — the shared floor of every examined man, deployed at reading speed. When the god calls his rivals 罔象, "phantom-shapes," he is not inventing an insult; he is lifting the term from the passage in which Confucius himself sorts anomalies by category. He does not argue with the frauds. He files them, in the classifying idiom of the sage, and the filing is itself the demonstration that he — unlike them — has the floor. The document is the examination answer. Its density is not ornament; it is the credential, laid on the page for inspection.

What the rubric is actually protecting

Now the market frame, which the text supplies itself and which resolves what the rubric is for. Twice the Sāi Bàng Wén names the harm that frauds do, and both times it is the same harm, and it is not the harm a Western reader expects. The counterfeits do not damn souls or deceive the pious into error. They "bring shame upon the god," 反為神羞; their trade is "the god's disgrace," 惟神之恥. But 恥 and 羞 here are not the anguish of a violated conscience. The harm is to standing — to the value of the genuine article in a market now flooded with fakes.

To feel why that harm is grave rather than merely awkward, one has to hold together three things a modern reader keeps in separate drawers: money, family, and the sacred. In the world the text was written into they were a single circuit. A shrine's custom is its income; its income feeds the living who tend it and funds the offerings to the dead who found it; and the offerings are what keep the ancestors fed in the other world and the lineage continuous in this one. Draw the custom away and you do not merely dent a revenue line — you starve the ancestors and break the descent. A rival who diverts the shrine's clients is therefore not a competitor in the mild modern sense. He is coming between a household and its dead. That is why the injury the text keeps naming is to the god's standing: standing is the asset on which the whole circuit — income, offerings, ancestors, continuity — depends. Debase it and every link downstream is threatened at once.

And the frame is explicitly commercial from end to end. The god's virtue is 無私, without private interest; the arch-fraud is the 壟斷賤夫, the base fellow of Mencius who climbs the high bank to corner the market — the passage, every examined reader knew, that Mencius offers as the origin of commercial taxation itself: the grasping man whose transgression first made oversight necessary. The final charge against the frauds is that they "flaunt themselves and peddle themselves," 自衒自鬻 — 鬻 is literally to sell — set against the model healer of an earlier couplet who sold medicine for thirty years and never named two prices. Disinterested office at one pole, the cornered market at the other, and the whole rubric strung between them.

Read this way the text stops being a defense of religious truth and becomes what it is: a trade-standards document. The rubric is a quality-control specification, and the whole point of a specification is that it is procedural, checkable, and indifferent to sincerity. The god is not protecting the faithful from error. He is protecting a house from counterfeiters. And it is worth being precise that this is not a cynical reading smuggled in against the grain of a pious text — nothing here is furtive, and there is nothing to expose. The enterprise is the piety. Feeding the ancestors is the business, and the business is the devotion; the two were never separate, so there is no seam between them to pry open. A cult is a going concern and a going concern is how the dead are served. A rival who can be consulted more cheaply threatens the custom, and the custom is the offerings, and the offerings are the ancestors — so he is dealt with accordingly. The stakes are not moral in our sense and they are not merely financial in our sense. They are both at once, because in this circuit those were the same thing, and in a world where cults lived or died by their licenses they were entirely, lethally real.

The engine underneath: the counterfeit works

This is the point on which everything turns, and it is easy to miss because it is never stated — it cannot be stated, since stating it would concede the case. The Sāi Bàng Wén would be unnecessary if the frauds failed. You do not draft a standards document, name your rivals by six marks, and sentence them to calamity by men and execution by ghosts because their planchettes produce nothing. You do it because their planchettes work — because the client who consults the cheaper, unlicensed medium down the lane gets an answer, and finds it good, and does not come back.

The whole apparatus of the third chapter — the rubric, the credential density, the licensing trail it supports — is the response of a franchise to open competition in a market where the product is, at the point of sale, indistinguishable. The technology is cheap. The format is copyable. And the result, to the customer, looks the same whether the operator has the floor or not. In that market the only move available to the established house is to relocate the question from did it work — where it cannot win, because the counterfeit works too — to who is qualified to have done it, where its command of the canon is decisive. The rubric is that relocation, executed in the god's own hand.

Which is why the sentence at the text's turn is so severe. "Men will bring calamity, and ghosts will carry out the execution," 人禍鬼誅. One does not answer an ineffective rival with the full jurisdiction of two worlds. The ferocity is not temper and it is not zealotry; it is a precise measure of how contested the ground is. You call down both courts only on a competitor who is winning customers. And the two courts are exactly right for the crime: a rival who diverts the shrine's custom offends against the living who depend on that income and against the dead whom the income feeds, so he is answerable in the human court and the ghostly one alike. The verdict matches the circuit it protects. The severity of the boundary-policing is the clearest evidence we have that the boundary was under real pressure — that the counterfeits were not failing, but succeeding, and had to be stopped by an authority the market could not confer and only the canon could.

The title asks whether the gods can be deceived, and the text answers, with its whole structure, that they cannot — because the accounts must balance, and an operator who lets them be gamed is a threat to the house. But the world the Sāi Bàng Wén was written into kept returning a different verdict, one the text can never quote and exists only to suppress: that it did not matter who held the stylus, because the demon left either way.

The reader who comes to this text expecting a religion, and finds instead a business, may feel he has caught the cult out. But there is no exposure here, because there was never anything hidden. The business was the religion. The bookkeeping was the devotion. What looks, to a gaze trained on the separation of the sacred from the commercial, like the unmasking of piety as racket, is only the discovery that in this world piety and enterprise were a single act — that to defend the shrine's income was to defend the ancestors, and to run the accounts was to keep faith with the dead. The Composition to Stop Slander is the licensed practitioner's refusal of the market's verdict — an attempt to reinstate, by the authority of the classics, a distinction that the market, judging by results, had already dissolved. That the distinction was also, and without contradiction, a matter of who got paid is not the scandal beneath the piety. It is the piety, seen whole.

Part SixThe Rhapsody on Not Being a God

The charge is in the title.

非神 fēi shén. Not a god. Two graphs naming what someone had said about the Perfected Lord of Áo Fēng, set at the head of his reply. The word 非 is the philosopher's against — the 非 of 墨子's 非攻, Against Aggression, and 非儒, Against the Confucians. So the title is at once the accusation and its refutation: Not-a-God, and Against "Not-a-God." The rhapsody carries the charge in its own name and answers it in the same breath.

And it answers in 賦 — the rhapsody, and specifically the 楚辭-descended southern rhapsody of first-person lament and self-vindication that goes back to 屈原, the 兮-cadenced line of the wronged and exiled speaker pleading his own case. This is a choice, and it is the section's spine. The form is the argument. To answer the charge of not being a god, the apparatus writes in the register the corpus reserves for divine authorship — the revealed genre, the god's own hand on the brush. Accepting the genre concedes the premise. Before a word of content, the mode of the reply has already asserted what the reply is trying to prove.

Descent instead of evidence

The rhapsody opens in cosmogony. 女媧 breaks the turtle's feet to prop up the sky; the primordial 渾淪 first parts at Song and Tai; the Five Peaks take their stations. Then it narrows to genealogy — the speaker's own line, from 下邳 Xià Pī to 采石 Cǎi Shí to 鱉峰 Biē Fēng, the Turtle Peak that is the cult's own ground. Then canonical citation, all the way down.

This is worth pausing on, because the method is the point. The apparatus faces, about itself, exactly the kind of charge it elsewhere levels at others: an ontological one. 非神 — you are not what you claim to be — is the same accusation, run in reverse, that the dispatches level at a frog or a drought-demon. And the rhapsody meets it the way the drought dispatch meets its target: not with evidence, not with argument, but with descent and record. The drought-demon is established as real by genealogy — descent from 炎帝, four canonical appearances across the classics — because a drought-demon can be neither produced in court nor caught in a net. The god answering the 非神 charge is in the same position, and reaches for the same instrument. He cannot be produced. So he recites where he comes from and cites who has spoken of it. Existence as a matter of record, because record is the only register available to an entity that cannot be exhibited.

Who decides

One couplet in the opening movement asks a question that looks metaphysical and is really jurisdictional:

遂右之傳而不證兮，誰謂水龍而木夔。

Suì yòu zhī chuán ér bù zhèng xī, shuí wèi shuǐ lóng ér mù kuí.

The transmission passed to the right and went unproven — who says

water-dragon and wood-kuí?

誰謂 — who says — reads at first as rhetorical scoffing. But set against the whole corpus's preoccupation, it is a question about authority: not do these marginal creatures exist but who holds the standing to rule an entity into the record — to say, with the force of the classifier, this is a water-dragon, that is a wood-kuí. It is the same operation as the Guoyu's Confucius sorting anomalies by their element, the same operation the Text to Stop Slander turns on its rivals. The rhapsody hands the skeptic's question back before he has finished asking it: you ask whether I am real; I ask by whose authority the real is named at all.

The borrowed brush

The middle sections turn from cosmogony to combat, and here the apparatus aims the skeptic's own canon back at him. The doubters are 井 jǐng — the well-frog of 莊子·秋水 who cannot conceive the sea — and 管窺, tube-peerers who mistake a glimpse for the whole sky. The choice of 秋水 is not casual: it is the canonical Chinese text on the limits of perspective, the standard resource of the epistemological skeptic, and the apparatus takes it and aims it back — the rationalist's doubt is not clear sight but a small vessel's view of a large thing.⁷

And one couplet in this stretch has usually been read too narrowly:

假毛錐子以為言兮，顛倒而妍媸。

Jiǎ Máo Zhuī Zǐ yǐ wéi yán xī, diān dǎo ér yán chī.

Borrowing Master Brush-Awl to make their speech — they invert the

beautiful and the ugly.

毛錐子 Máo Zhuī Zǐ, "Master Brush-Awl," is the writing brush personified, with a reach back to Han Yu's 毛穎傳, the mock-biography that gives a brush a life and a surname.⁸ The natural gloss is that the slanderers wield a pen — that this is about polemic. But 假 jiǎ is not a neutral verb. It means borrow, and it means counterfeit — to take on falsely — and the same verb governs the central accusation in the companion text two slots away, where fraudulent mediums 假以狐鳴, counterfeit the fox-cry, the Shiji's locus classicus for a manufactured supernatural sign (Chen Sheng's man hidden by the shrine at night, calling in an animal's voice to fake an omen).

Read against that, the couplet is not describing men writing polemic. It is describing men operating a brush that is not theirs to operate — speaking in a borrowed voice, through an instrument meant to be moved by someone else. Which is to say: rival planchettes. The accusation the god levels is the very one he is himself defending against, aimed the other way.

The rationalist's own outcome

The closing sections widen from attack to a longer view, and then set one canonical case at the center:

阮瞻之辯是非兮，骨囪化為塵泥。

Ruǎn Zhān zhī biàn shì fēi xī, gǔ xìn huà wéi chén ní.

Ruǎn Zhān argued the case of are-they versus are-they-not — and his bones

and skull turned to dust and mud.

阮瞻 Ruǎn Zhān, of the Bamboo Grove, is the Shishuo xinyu's famous denier of ghosts: the man who argued the rationalist position at length, was visited by one, and died shortly after.⁹ The apparatus does not offer this as proof. It offers it as the rationalist's own outcome, drawn from the rationalist's own canon of anecdote. The argument is not ghosts exist, and here is the demonstration. It is the man who argued they do not is now dust, and the record of his arguing is itself a ghost story. The skeptic is answered out of his own bookshelf.

The refusal

The coda states the principle — 天道之福善禍淫, Heaven's Way blesses the good and brings disaster on the licentious, automatic as a shadow following a sundial's gnomon — dismisses the doubters as 狐疑, fox-doubters hesitating at every step of the frozen river without ever crossing, and then ends like this:

敬神如在左右兮，予嘿嘿而何辭。

Jìng shén rú zài zuǒ yòu xī, yú mò mò ér hé cí.

Reverencing the spirits as if present to left and right — I, in silence —

and what is there to say?

敬神如在 is the Analects: 祭如在，祭神如神在 — sacrifice as if they were present, sacrifice to the spirits as if the spirits were present. Confucius's formula is famously non-committal about metaphysics and entirely committed about posture: it prescribes a disposition, not a belief. And then, silence.

This is the essay's crux, and it should be stated plainly. The rhapsody does not answer the question it was written to answer. Asked to prove that it is a god, it recites the cosmogony, gives its genealogy, dismantles its accusers on borrowed philosophical ground, notes what became of the last man to argue the negative — and then declines to continue on the terms offered. The refusal is not evasion. It is a claim about what kind of question this is. Is this entity metaphysically real is a question the rhapsody treats as badly formed, and it answers instead the question of proper conduct in the presence of what cannot be verified either way: reverence as if present, silence as to the rest.

Paul Veyne asked whether the Greeks believed in their myths and found the question harder than it looks, because belief runs in more than one register at once.¹⁰ The Rhapsody on Not Being a God is a document from inside that difficulty, written by an apparatus that understood the difficulty perfectly well and declined to resolve it in the skeptic's favor or its own.

What the two texts share

One further thing, which will matter for the final section. The rhapsody and its prose companion are written in one another's vocabulary. Inversion (顛倒) appears in both. The verb of counterfeiting (假) carries the central accusation in both. The rhapsody closes on silence — 嘿嘿 — and the same word for silence sits in the verse-charge of the Text to Stop Slander.

And one word threads them at a deeper level. In the rhapsody the speaker says 吾剡剡於羅池 — I shone forth, whetted and glittering, at Luo Chi — naming his own manifestation. In the companion text the same word, 剡剡, describes the god at the moment he becomes the brush. Sharpened; whetted; glittering — a blade-word and a brush-tip word, used at once for the appearing and for the instrument. In this corpus, manifestation and brush-point are named with one term. Which is where the last section begins.

Part SevenOne Repertoire, Two Deployments

The Rhapsody on Not Being a God ends by refusing to answer. Asked to prove he is a god, the god falls silent — 予嘿嘿而何辭, I, in silence, and what is there to say? It is the right ending for that text, and it leaves this essay with the one question the corpus will not settle for us. If the god will not say whether he is real, we are left only with what is on the page. So it is worth asking, at the end, what the page can tell us that the god declined to — not about whether a spirit held the brush, which is unrecoverable, but about who did.

Consider what the two texts we have read actually required of whoever produced them. The Composition to Stop Slander convicts its rivals by lifting 罔象 out of the Guoyu, where Confucius sorts anomalies by their element; it invokes the Yijing's hexagram of Abundance to describe an empty room; it reaches for the Mencius parable of the man who corners the market, and trusts the reader to supply the rest. The Rhapsody answers its own charge in the rhapsodic register of the Chu ci, the form of Qu Yuan's self-vindication, and closes on a silence borrowed from the same tradition. Neither text explains its allusions. Both assume a reader who already has the Analects, the Odes, the Mencius, and the Songs of Chu by heart — because both were produced by someone who did.

It is tempting to divide this equipment in two. There is the expelling work — the crocodile driven out, the drought-demon dismissed, the pestilence warded — which reaches into the classics for their machinery against the noxious: the 蟈氏 of the Zhou li with its published recipe for killing frogs, the bear-demon of the Zuo zhuan, the Cloud-Han ode against drought. And there is the arguing work — the rival unmasked, the counterfeit named, the slander stopped — which reaches into the same shelf for its weapons of category and precedent. It looks like two kits: a ritualist's and a polemicist's.

But it is one kit. The Zhou li and the Mencius sit on the same shelf, in the same memory, learned in the same schoolroom for the same examinations. The man who expels a crocodile with a canonical precedent and the man who expels a rival with a canonical precedent are not two specialists sharing a library. They are one specialist doing one kind of thing twice. And the kind of thing is the same in both cases, which is what the reading in this essay has been circling: a threat is named, its category is fixed, and on the strength of that fixing it is expelled. The crocodile and the counterfeit medium are, to this apparatus, the same sort of problem — a predator on the ground the god is appointed to protect — and they receive the same treatment, classified out of the canon and driven off. The ritualist's gesture and the polemicist's gesture are one gesture. The kit is single because the operation is single.

Which tells us something about the hand on the stylus that the god's silence would not. Whoever was producing these texts had the full equipment of an examined man — not the partial, practical literacy of a village healer who has memorized a handful of efficacious phrases, but the deep, ramified command of the canon that the civil examinations were designed to test and certify. The allusions are not decoration and they are not display for its own sake; they are the working tools of the trade, reached for at speed and deployed without explanation because the audience shared the training. The planchette at Áo Fēng was operated, whatever else was or was not true of it, by someone who could have sat the examinations — and who may well have.

That is as far as the texts will carry us, and it is worth being honest about where it stops. To say the operator had an examined man's command of the canon is not to name him, or to say how many hands passed the stylus across three centuries, or to settle whether the learning on the page belonged to the medium in trance, the recorder who copied the sand, or the editor who gathered and polished the whole for print. The apparatus ran to six functionaries at full strength; the text we read is the product of all of them and attributable to none in particular. The canon on the page is certain. Its custody is not.

And so the questions this essay leaves open may be worth more than the ones it has closed. If the ritualist's expulsions and the polemicist's arguments are one operation performed with one training, then the boundary we draw between Daoist ritual and Confucian letters — the boundary that files the crocodile text under religion and the Mencius citation under scholarship — may be our boundary and not theirs, a shelving convention imposed on a body of work that did not recognize it. If the same hand that cleared a demon from the water cleared a rival from the market, using the same canon for both, then the question of who held the stylus opens onto a larger one: how much of what we sort into religion, in this world, was the ordinary work of the lettered class, done in the sacred register because that is where the work happened to be needed. The Hànzǎo does not answer that. What it does, and what the third chapter does most sharply, is show us the question the whole corpus was built to manage — and it is not, in the end, whether the gods are real. It is who has the authority to say.

That is the thread that runs from the title onward. Can the gods be deceived is a question about verification: whether the channel can be gamed, whether a false signal can pass as true, and — the moment the first two prove unanswerable — who is qualified to tell the difference. Every section has been the same question in another guise. Efficacy leaks into text: how does the numinous become legible at all? The technology spreads: whose planchette, by what standing? The rubric inspects the operator and never the message, because the message cannot be inspected and the operator can. The rhapsody asks who says water-dragon and wood-kuí — who holds the power to rule a thing into the record — and then, asked to prove itself, refuses, and prescribes reverence in place of proof. These are not separate arguments. They are one argument about authority, conducted by a tradition that had understood, six centuries before Paul Veyne framed it for the Greeks, that is it real is the wrong question — too crude to catch how belief actually operates — and had replaced it with the questions that can be worked: who is qualified to certify, by what procedure, and what is owed to a truth that cannot be produced in court.

This is what makes the third chapter more than a curiosity of a minor Fujian cult. Faced with the one charge it could never rebut — you cannot prove you are a god — the apparatus did not attempt the impossible proof. It built a credential. It wrote the rubric that says who may be believed, and the rhapsody that performs the qualification, and it filed them as the standing papers of an authority that no plaque could confer and only the canon could. It answered are you real with here is who is entitled to ask, and here is who is entitled to say. The gods cannot be deceived because deception is beside the point; what is at stake was never the honesty of the report but the standing of the reporter.

Which is why a reader who closes this book wondering where the line between an exorcism and an essay was ever supposed to fall has understood exactly what the third chapter was quietly showing all along: that the line was never between the sacred and the secular, or the real and the fabricated, but between those with the authority to name the world and those without it — and that the god of Áo Fēng, holding the brush, had appointed himself to the naming.

Notes

Notes

1. 徐仙翰藻, HY 1456 (= CT 1456), 14 juǎn, in the 萬曆續道藏 Wànlì Xù Dàozàng. Compiled 1305 (大德九年) by 陳夢根. See Judith Boltz, A Survey of Taoist Literature: Tenth to Seventeenth Centuries (Berkeley, 1987), §19. ↩
2. 徐仙真錄 Xú Xiān Zhēn Lù, HY 1458, preface, dated 大明永樂甲辰 (1424). The full clause lists the genres: 並真君降筆所為古賦、記序、詩歌、籤兆. ↩
3. 紀昀, 《閱微草堂筆記》 — 遇能書者則書工，遇能詩者即詩工，遇全不能詩能書者，則雖成篇而遲鈍. ↩
4. On the Song title-granting system and the 正祀 / 淫祀 distinction, see Valerie Hansen, Changing Gods in Medieval China, 1127–1276 (Princeton, 1990). ↩
5. The Hànzǎo is well described in the scholarship, though at a different level of focus. Kristofer Schipper identified the whole collection as spirit-written: "All the pieces contained in this collection were written by the immortal Xu brothers themselves through spirit-writing" (Schipper and Verellen, The Taoist Canon [Chicago, 2004], 1211). Judith Boltz catalogues the corpus in her Survey of Taoist Literature (Berkeley, 1987), treating the cult of Vast Mercy (洪恩 Hóng'ēn) across three sections and describing chapters 3–6 as "a diverse selection of fu rhapsodies, sung lauds, hsü prefaces, and tsan encomia," from which she singles out the juǎn 4 dispatches against malaria and drought (§IV.19, pp. 195–97). Edward Davis treats the cult's spirit-writing and pointedly multi-traditional character ("Arms and the Dao, 2: The Xu Brothers in Tea Country," in Kohn and Roth, eds., Daoist Identity [Honolulu, 2002], 149–64), and Zhang Chaoran 張超然 has analyzed its techniques of Daoicization. These treatments describe the corpus at the level of attribution, cult history, and genre. So far as I have found, none isolates the argument of the Sai Bang Wen in particular and reads it as a professional rubric — a working standards-document for distinguishing authentic spirit-writing from fraudulent, issued by the deity who sets the standard. That reading is offered here. Recent work by the Columbia dissertation The Descending Words: Cult and Culture of Spirit-Writing in China (900–1400) (2025) examines Song–Yuan spirit-writing and questions of textual authenticity, though not, in its abstract, the Xu corpus specifically. ↩
6. The title is romanized here Sāi Bàng Wén, following the reading of 塞 as sāi, "to stop up, to plug." The graph also reads sè in the same sense of blocking or damming — the more common reading for the abstract verb, as in 塞責 or the sibling title 塞井文 in juǎn 4 — and a reader will meet the title spelled Sè Bàng Wén elsewhere. The two are readings of one character in one meaning: to stop the slander up, as one stops a well or a gate. (The third reading, sài, "frontier fortress," is unrelated.) ↩
7. 莊子·秋水 — the Autumn Floods chapter of the Zhuangzi, whose opening parable of the frog in the collapsed well (井鼃) who cannot be told of the sea, and the standard image of measuring the sky through a tube (管窺), are the classical Chinese loci for the limits of a narrow perspective. ↩
8. 毛穎傳, "Biography of Fur-Point," Han Yu's mock-biography personifying the writing brush as a man with a surname and a career. 毛錐子, "Master Brush-Awl," is a later term of the same personifying habit. ↩
9. The anecdote is 《世說新語》; Ruan Zhan 阮瞻, known for maintaining that ghosts do not exist (無鬼論), is confronted by a visitor who argues the point, reveals himself to be a ghost, and vanishes, after which Ruan sickens and dies. ↩
10. Paul Veyne, Les Grecs ont-ils cru à leurs mythes? (Paris, 1983); English: Did the Greeks Believe in Their Myths? An Essay on the Constitutive Imagination, trans. Paula Wissing (Chicago, 1988). ↩