If you want to use a stratagem in an argument and cite Sun Tzu for it, the short answer is that you usually cannot. The Thirty-Six Stratagems and The Art of War are separate books, roughly a thousand years apart, and the collection is not a digest of the treatise. So when you see a stratagem filed under a chapter — on our own Strategy Map, or in any of the dozens of charts that circulate online — that link is a reading aid, not a lineage.
A reading aid is only as good as its sourcing, and these divide into three kinds that are worth very different amounts.
Three kinds of link, and only one of them is a citation
Any pairing of a stratagem to a chapter falls into one of them.
The treatise uses the stratagem's own words. This is rare and it is the only case where "chapter 7 says so" is literally true. 以逸待劳, wait at leisure for the exhausted enemy, appears in chapter 7 (军争, Maneuvering) as 以佚待勞 — the same four characters, one of them written with a variant graph — inside a passage on husbanding strength. 反间计, the double agent, is chapter 13's own vocabulary: 反間者，因其敵間而用之, the doubled agent listed as one of the five classes of spy. Cite those two and you are quoting, not interpreting.
The treatise supplies the image. 上屋抽梯, lure onto the roof and remove the ladder, is not a phrase in The Art of War — but chapter 11 contains 如登高而去其梯, "like one who climbs high and kicks the ladder away", describing what a general does to his own army when he commits it past retreat. The picture is identical; the stratagem later pointed it at the enemy instead. That is a strong link and still not a derivation.
Most links only echo the principle. This is the largest group by far. 釜底抽薪, take the firewood from under the pot, sits with chapter 2 because that chapter argues the economics of supply — 智將務食於敵, the wise general feeds off the enemy. Nobody in chapter 2 mentions a pot. The link argues that the move belongs to that chapter's logic. That is a useful connection, and it is not a citation.
One of the thirty-six has no home at all
借尸还魂 — borrow a corpse to resurrect the soul — is linked to no chapter, and that is a decision rather than an oversight. The stratagem is about manufacturing legitimacy by reviving a dead name, a lapsed title, or a powerless heir and pouring your own purpose into the vessel. Its origin story is the enthroning of a shepherd boy as King Huai of Chu in 208 BCE to give an anti-Qin coalition a banner.
The treatise has nothing to say about that. It is a book about a commander in the field, and the manufacture of political legitimacy is outside its frame. Every mapping we could construct for it rested on a resemblance rather than a line, so the map shows it as unlinked and says why. A chart that gives all thirty-six a chapter has, somewhere, made one up.
What the audit found in our own map
The claim above — that every link rests on a named line — is one we were not entitled to make until this week.
The map's data file carried a note saying the links were "doctrinal, not decorative: each rests on a line of the chapter it sits in, recorded in _related_rationale." Reading the file end to end against the thirteen chapters, that was true of 17 links and false of 26. The 26 undocumented ones were, exactly, the first two entries in every chapter's list — thirteen chapters, two apiece — which is the shape a list takes when it is filled to a quota rather than by reading.
The useful part is what happened when we went to ground them. All 26 held. Every one turned out to have a real line behind it; not a single link had to be deleted. The pairings themselves were not the defect. The problem was that we had a record of the reasoning for some, and a sentence claiming it for all. That is a smaller failure than a wrong map and a more insidious one, because a wrong map can be argued with and an unsourced one cannot.
Two links were also added, to settle a contradiction between two of our own pages. Two of the stratagems' own pages cite a chapter directly: 以逸待劳 points to chapter 7, and 趁火打劫 to chapter 1. The crosswalk had 以逸待劳 under chapters 2, 4, 8 and 10 — every plausible chapter except the one that contains its name — and 趁火打劫 under chapter 12 alone. Both pages were right and the map disagreed with both, on adjacent pages of the same site.
The map now carries 45 links across 13 chapters, covering 35 of the 36, and each one displays the line it rests on. A test fails the build if a link has no line, if a line has no link, if a line contains no quoted source text, or if a stratagem page names a chapter the crosswalk does not link it to.
The weakest link, named
One pairing survived the audit on sufferance. 美人计, the beauty trap, is filed under chapter 13 because that chapter treats the recruitment of an enemy's own officials as a discipline — 內間者，因其官人而用之 — and the beauty trap is a method of doing precisely that. It is the loosest fit on the map. We kept it because it helps a reader place the move, and we labelled it in the data as a reading aid, not a derivation.
Naming your weakest evidence is cheap and it is the whole difference between a map and a decoration. If we had quietly dropped it, the map would look tidier and tell you less.
How to use one of these links without overclaiming
Three rules cover using these links without overclaiming.
Say what you are citing. "Chapter 7 uses this phrase" and "chapter 2 supports this reasoning" are different claims, and the second one is fine as long as you do not dress it as the first.
Check that the line exists. Every link on our map shows its source text; if a chart you are reading does not, treat the pairing as somebody's opinion — which it may well be, and a defensible one.
Do not let a link launder a stratagem's history. Whether the treatise supports a move is a separate question from whether its attached story actually happened. We counted that separately; the answers do not line up. The empty fort has an excellent chapter-6 grounding and a famous telling that a fifth-century historian rejected.
The full crosswalk, all 45 links with their source lines, is on the Strategy Map, which also shows the thirty-six arranged by their own six sets. If you are working the other direction — you can see what someone is doing and want to know what to do about it — the Counter-Stratagem Finder starts from the signals rather than the name.
