A general has marched his army up to an enemy city and now faces the oldest hard question in siege warfare: can I take this place, or will it break my army against its walls? He cannot see inside. He does not know if the garrison is starving or fresh, terrified or defiant, loyal or ready to defect. So — according to a chapter of the Sung-dynasty Hu-ling Ching called "Cloud Ch'i Above Cities" — he does something that looks, to us, absurd. He stops looking at the walls and looks at the sky above them. The shape and color of the cloud-vapor hanging over the city is supposed to tell him everything the walls hide: whether the men inside want to surrender, want to sue for peace, are sick, are about to sally out and fight, or are protected by Heaven itself.1
That is the wager of this doctrine. A besieged city, packed with frightened men, supposedly broadcasts its hidden condition upward as a readable vapor — and the besieger who can decode that vapor knows the outcome before he commits a single ladder to the wall. The chapter is, in plain terms, a go / no-go decision table for sieges, indexed not on intelligence from inside the city but on the clouds over its rooftops.
The "Cloud Ch'i Above Cities" chapter continues an old project: Sun Pin had already tried to characterize cities as susceptible to attack or not, and the Hu-ling Ching says it is echoing "the T'ai Kung's early ch'i indices."2 So this is not a one-off curiosity — it is one late, elaborated entry in a long tradition of scoring a walled target. What's distinctive is the medium. Earlier city-evaluation weighed walls, supplies, terrain, and the defenders' spirit. This chapter weighs the atmosphere.
The opening instruction sets the frame for both attacker and defender: "Whether you want to advance your army and attack a city or the enemy is coming forth to besiege you, you should investigate the auspiciousness and balefulness of the clouds and ch'i."3 The doctrine cuts both ways — you read the sky over the city you mean to take, and you read the sky over your own position when you are the one besieged.
What follows is a dense lookup grammar, each entry pairing a cloud-form with a verdict and, usually, an order. A sample of the readings:
Surrender, peace, plague, doom-with-a-back-door: each state of the city has its sky-sign.
The grammar runs on three variables — the shape of the ch'i, its color, and its motion relative to the city walls — and it converts combinations of these into a verdict about the men inside. Shape carries human meaning: a vapor arrayed "like men kneeling or crouching... with heads bowed and hands folded" reads as submission, because it looks like submission.8 The bowed-head, folded-hand cloud is read exactly as a bowing, surrendering man would be read. This is sympathetic reading — the sky is taken to mimic the posture of the souls beneath it.
Color carries fortune: yellow over white is "excellent," signaling a wish for peace; ch'i "like dead ashes," grey and lifeless, signals a garrison that is itself dying of sickness.9 The dead-ashes reading is the cleanest example of the doctrine's logic — the color of death in the sky is read as death in the city. Lifeless vapor, lifeless men.
Motion across the wall-line carries direction of fortune: ch'i that goes out of the city means the populace "wants to sneak out and disperse and lacks the will to fight" — so attack quickly.10 Ch'i that "mistily surrounds a city but doesn't enter" means troops outside will not succeed in getting in.11 The vapor crossing the wall, or failing to, maps onto bodies crossing the wall, or failing to.
Folded into the same table are readings that protect the attacker from himself. Several signs are warnings: green ch'i "like a cow's head butting a man" means the people within cannot be slain; yellow ch'i from the eastern quarter indicates "Heaven's Ax" and the city "cannot be attacked" — "anyone who attacks it will suffer great misfortune."12 And ch'i "like chickens, pheasants, or running rabbits" is not about the city at all but about you: brigands are coming to attack your forces, prepare urgently.13 The table reads threats to the besieger as readily as weaknesses in the besieged.
Under the cosmology, the thin grain of real observation is the same as in the broader ch'i tradition: a real city under siege does throw smoke, cooking-vapor, and the grey pall of disease and burning into the air, and a watcher could sometimes infer something true from it. The doctrine takes that grain and inflates it into a complete sky-language where a sword-shaped cloud promises subversion in exactly eighty-one days.
This page is the siege specialization of Ch'i Cloud-Reading Doctrine. The parent doctrine claims that any body of a hundred-plus men broadcasts its fate as a visible vapor; this page points that general claim at the single hardest intelligence target in pre-modern war — the sealed interior of a walled city. Where the parent reads armies in the open field, this reads the one force you fundamentally cannot see.14
It hands City Evaluation and Targeting a second, parallel scoring system. City-evaluation scores a target on tangible indices — walls, stores, terrain, defenders' morale, leadership. This chapter scores the same target on an atmospheric index. The two sit side by side in the corpus as alternative answers to the identical question — is this city takeable? — and their coexistence is itself the interesting datum: the tradition kept a tangible scoring method and a portent scoring method running in parallel without forcing a choice.
And it connects directly to Sun Pin's Male and Female Cities. The source names Sun Pin's project — sorting cities into those susceptible to attack and those not — as the lineage this chapter "continues."15 Sun Pin sorted cities by terrain and configuration; the Hu-ling Ching sorts them by the cloud overhead. Same binary (takeable / untakeable), different oracle.
The richest single entry is the white sword-cloud. "If the white ch'i gleams like a long sword more than a hundred staffs long that rises up above the enemy and stretches over the city, anyone who attacks will suffer disaster. The city cannot be slain, but within eighty-one days you can effect a subversive response."16 Sit with the structure of this reading, because it is more sophisticated than the simple surrender/slaughter binaries around it.
It does two things at once. First, it forbids the assault — a sword in the sky over the city means a frontal attack brings disaster, so the table tells the eager general no. Second, it does not leave him with nothing. It re-routes him: force won't work, but subversion will, and it even stamps a deadline — eighty-one days. So the same sign that closes one door opens another, and the doctrine pivots the besieger from the wall to the agent, from storming to turning men inside.
The precise number is the tell. Eighty-one days is not an observation; nothing in a cloud could yield that figure. It is the grammar asserting its own completeness — converting a vague warning into a confident, dated operational plan. This is exactly the seductive danger of the whole apparatus: it doesn't merely caution, it prescribes, with a specificity no cloud could ground. A general who trusts the sword-cloud will call off a sound assault and sink eighty-one days into recruiting traitors who may not exist, on the authority of a vapor. The reading is operationally complete and evidentially empty, and those two facts travel together.
You are the besieging general, and it is the grey hour after dawn. Your siege lines have been set for days. Before you give the order that will spend men's lives against the wall, you walk to the rise where you can see the whole city laid out below — and you lift your eyes above its rooftops.
The ch'i over the eastern gate hangs low and the color of cold ash. You know that reading: dead ashes, the garrison sick, the city can be slaughtered.17 Your pulse quickens — this is the sign that says go. But you make yourself look at the rest of the sky, because the table punishes the man who reads only what he wants. Over the citadel, a pale streak rises and lengthens, gleaming, narrow as a blade laid across the heavens. A white sword-cloud. The other reading lands cold against the first: attack here and suffer disaster; do not storm; turn men instead, within eighty-one days.18
So you hold. You do not throw your ladders against the citadel. You call for your handler of agents and you start counting toward eighty-one days, because the sky has told you the wall will not fall to force but the city will fall to subversion. You have read the enemy not by getting a man inside, but by reading the air the way the Hu-ling Ching taught — and now your whole campaign bends to a cloud.
The doctrine fails the moment the general lets the sky overrule the situation on the ground. A sword-cloud that calls off a well-supplied, well-timed assault against a genuinely weak wall has cost an army its best chance for nothing. The warning sign is a commander who can no longer tell you why he isn't attacking except by pointing upward — whose siege decisions are anchored to vapor rather than to walls, stores, and men.
The subtler failure is the doctrine's false precision. "Eighty-one days," "for the next seven days," "more than three days" — these dated, numbered prescriptions feel like hard intelligence and are nothing of the kind. A practitioner who has begun trusting the numbers the clouds supply has crossed from observation into oracle. And because the table contains a sign for every outcome — surrender, peace, plague, doom, ambush of the besieger himself — it can always be made to speak, which means it can always be made to justify the decision the general already wanted. When the cloud is doing the deciding and the wall is just the thing the cloud happens to hang over, the siege is being run by the weather.
The chapter closes with a line that betrays its own anxiety: "Now the ch'i for victory or defeat above a city are such. If it indicates victory for the enemy, you cannot attack, but if defeat for them, you should attack... All cloud ch'i manifests the mind of Heaven and Earth, so how can you not be cautious?"19 The doctrine presents itself as Heaven's own judgment — and yet its first practical instruction is the entirely mundane don't attack a city that's going to beat you, do attack one that's going to lose. Strip the clouds and that's just sound generalship. The portent-language wraps a banal truth (attack the weak, spare the strong) in the authority of Heaven, and the chapter never quite admits which is doing the work.
The unresolved tension is the gap between the readings that have a grain of physical sense and the readings that have none. Dead-ashes ch'i over a plague-stricken city could be real — disease and death foul the air. A sword-shaped cloud promising subversion in eighty-one days could not be. The chapter files both as equal entries in one table, with no marker separating observation from omen. Whether this grammar grew from a real besieger's eye for smoke and pall and then accreted the dragons and swords, or was a portent-system from the start, the text does not say — and Sawyer, presenting it as a continuation of Sun Pin and the T'ai Kung across many centuries, leaves that continuity claim resting on the texts' own back-references rather than on independent confirmation.
Sawyer frames this chapter as one node in a long, self-citing lineage — Sun Pin to the T'ai Kung's "early ch'i indices" to the late Sung Hu-ling Ching — and that very framing is where his sources pull against caution.20 The chapter claims descent from antiquity; Sawyer reports the claim, and a reader has to notice that a late text asserting it continues "the T'ai Kung's early indices" is exactly the kind of antiquity-projection the corpus habitually performs. The convergence Sawyer draws is real: across centuries, Chinese military culture kept asking is this city takeable? and kept generating answers. The friction is that this particular answer-method dresses ordinary siege prudence ("don't attack a city that will beat you") in Heaven's robes, and the chapter's own closing line — half operational rule, half cosmic invocation — shows it cannot fully decide whether it is reading clouds or reading sense. Sawyer preserves that ambivalence rather than resolving it, which is the honest move, but it leaves the historian to separate the siege-craft from the sky-craft on his own.
Handshake to City Evaluation and Targeting. These two pages are rival answers to one question, and laying them together exposes something neither shows alone. City-evaluation scores a target on auditable inputs: wall height, granary depth, terrain, the defenders' visible morale, the quality of their commander. This chapter scores the identical target on an unauditable input: the cloud above it. Both deliver the same verdict-type — takeable or not, attack or hold. The insight is that the corpus ran a falsifiable and an unfalsifiable scoring system in parallel and never demanded they agree, which tells you the intelligence culture valued having an answer over having a checkable one. When the auditable indices were ambiguous — and against a sealed city they usually were — the cloud-table filled the gap with confident, dated verdicts precisely where real information was thinnest. The danger lives exactly there: the portent method is most attractive in the situations where it is least anchored, because that is when the general most needs to be told what to do.
Handshake to Sun Pin's Male and Female Cities. Sun Pin sorted cities into "male" (strong, defensible, not to be attacked) and "female" (weak, takeable) by reading their terrain and configuration — the ground they stood on, the rivers and heights around them. This chapter inherits the same binary and the same goal but swaps the evidence base from earth to sky. Put side by side, they reveal the tradition migrating its city-classification from physical site to atmospheric sign — from something you could walk and measure to something you could only gaze at. The insight is that as the corpus elaborated across dynasties, the target of analysis (takeable city / untakeable city) stayed fixed while the medium of analysis drifted toward the unfalsifiable. Sun Pin's male/female reads a real feature of the world; the cloud-table reads a projection of the besieger's hopes. The continuity of the question across that drift is what makes the drift visible: same binary, ground beneath it dissolving into vapor.
Sharpest implication: The chapter is a complete go/no-go siege table whose verdicts grow most precise — dated to the day — exactly where its evidence is thinnest, so its false precision is at its worst against the very target (the sealed city) it exists to read.
Generative questions: