Paris, Café de la Régence, traditionally 1750. Kermur de Légal (White) v. Saint Brie (Black). 1–0, 7 moves, as usually published.

Document 1. What we know, and how

Every move below is the standard published score, and every move was replayed by code before this page went up: each move is legal, each check is a check, the only reply after the sixth move is the one shown, and the last move is mate. The score is held in the ClassicsForge bank on this site, which is the authority if any copy disagrees.

That is more certain than the game itself. Three things are open. The date is usually given as 1750, but research suggests the 1780s are more likely. Légal is said to have played this game giving the odds of a rook, playing without his a1 rook; it never moves in the published score, so the odds change nothing about the moves. And the move order may have been revised after the fact. The likely original, according to later research, had Black's knight already on c6, which would have given Black a refutation (see Document 5). Treat what follows as the cleanest version of a real idea, not a transcript.

Document 2. The score, with notes

1. e4 e5 2. Bc4. The bishop aims at f7, the one square in Black's camp defended only by the king. It is the target of the whole game.

2... d6 3. Nf3 Bg4. The decision the game turns on. Black pins the f3 knight to the queen before developing a single kingside piece. This is a relative pin: the knight is legally free to move, but moving it uncovers the queen. A relative pin is a statement about price, not a rule.

4. Nc3 g6. Black plans ...Bg7, which is slow. Count the position, not the material: White has three pieces developed and can castle next move; Black has one piece out, and everything Black is doing depends on that one piece holding the pin.

5. Nxe5! The pinned knight moves anyway. This is the move every beginner's rule forbids ("don't move a pinned piece", "don't give away your queen"), and it is the correct move, because the price of breaking the pin is lower than it looks.

Black has two captures. Consider both.

• 5...dxe5 6. Qxg4. White gave a knight (3) and got a pawn (1) and a bishop (3). White is a pawn up, the pin is gone, and Black's only developed piece is gone. This is the line the engine check calls "material, White at least +1", and it is why the knight is free to move: refusing the queen does not save Black, it only loses more slowly.
• 5...Bxd1? Black takes the queen. This is the move that gives the game its name.

6. Bxf7+ Ke7. Forced. The king cannot take the bishop (the e5 knight guards it), cannot go to d7 (the e5 knight again), and f8 is blocked by Black's own bishop. Ke7 is the only legal move.

7. Nd5#. Mate by three minor pieces. Check the escape squares yourself: d6 and f8 are blocked by Black's own pawn and bishop; d7 is covered by the e5 knight; e6 and e8 by the f7 bishop; f6 by the d5 knight; nothing can take on d5 because the d6 pawn shuts the black queen's file. The bishop that captured White's queen is on d1, as far from the action as a piece can be.

Document 3. The structure of a trap

The game has a shape that recurs far outside chess. Three conditions combine:

1. A belief that is usually true. "A pinned piece cannot move" is a good rule. Most of the time, moving a pinned piece really is bad.
2. A plan built on the belief. Black's whole development depends on the pin holding. When the belief fails, there is nothing else.
3. A visible prize that is actually a cost. The undefended queen is real. It is also the thing that pulls Black's attention, and Black's bishop, away from the square where the game is decided.

The useful question is not "is this a trap?", which you often cannot tell. It is "what happens if the rule I am relying on is false here?" That question can be asked every time, cheaply, and it is usually answered "nothing much". The one time it is not, it matters a great deal.

Document 4. Relative and absolute pins

A relative pin freezes a piece in front of something valuable. An absolute pin freezes a piece in front of the king, and then the pinned piece is not allowed to move at all. Black's 3...Bg4 is relative. That is the whole difference between this game and a sound attack: a relative pin is only as strong as the cost of breaking it. Here, the cost was a queen for a mate, which is no cost at all.

Compare the Opera Game, where Morphy's pins hold because he keeps adding pressure until breaking them costs more than keeping them. A pin is worth what you can keep pointing at it. Saint Brie pointed one bishop at a knight and nothing else.

Document 5. The probable original, and the flaw

Later research suggests the game was really played in a different order, roughly 1. e4 e5 2. Bc4 d6 3. Nf3 Nc6 4. Nc3 Bg4 5. Nxe5? Bxd1??. In that order Black's knight is already on c6, and after 5.Nxe5 Black can simply play 5...Nxe5, taking the knight with a piece instead of the queen. Black then wins a piece. The famous combination only works because Black was greedy.

The version in every book removes the knight from c6, so the combination works against any reply. That makes it a better lesson and a less accurate record. Both things are worth holding. A teaching game is often a cleaned-up game, and knowing that it has been cleaned up is part of reading it well.

Document 6. A note on the odds

Playing at odds was ordinary in eighteenth-century cafés. A stronger player removed a piece, often a rook or a knight, to make a game against a weaker one worth playing. Rook odds are a large handicap. If the story is true, it adds a second layer: Légal could afford to give away his queen because he was already playing a piece down and was used to winning on activity rather than material.

Document 7. How Black avoids it

Three habits keep Black out of this trap, and none of them requires seeing the combination in advance.

Develop knights before bishops. The old guideline is not a law, but it fits here. With ...Nc6 and ...Nf6 played first, the e5 pawn is defended by a piece, and 5.Nxe5 simply loses a knight. The combination needs e5 to be defended only by the d6 pawn, so that taking on e5 costs White nothing he cannot win back.

Do not pin what you cannot keep pinned. Before playing ...Bg4, ask what keeps the bishop there. If the answer is "the knight is pinned, so nothing can attack me", the pin is carrying the whole weight of your plan. Morphy's opponents made the same choice a century later in the Opera Game and lost the bishop one move after placing it.

Look at f7 every move. In the opening, f7 is guarded only by the king. Whenever a white bishop is on c4, any white piece that can also reach f7 is a threat. Here the e5 knight is the second attacker. Counting attackers and defenders on that one square would have warned Black in time.

None of these habits is clever. That is the point. Traps catch people who are thinking hard about the wrong thing, and they are defeated by small dull routines done every move.

Document 8. Why the trap survives

You would expect a trap this famous to stop working. It hasn't, for two reasons.

First, it can arise from several openings and several move orders, so a player who knows it by name may not recognise it when it appears in a slightly different shape. The pattern is the bishop on c4 eyeing f7, a knight that can reach e5 or g5, a second knight that can reach d5, and an opponent's bishop pinning a knight to the queen. Learn the pattern, not the moves. A pattern you can describe in a sentence will travel with you into positions you have never seen; a memorised sequence will not survive the first move your opponent plays differently. That is also why the annotation above stops to check every escape square rather than simply announcing mate. The check is the pattern.

Second, the bait is the queen, and very few people can look at a free queen calmly. That is a fact about players, not about chess.

Document 9. Your position

Set up the position after 4...g6. You are White. You are also, for a moment, the person who has to decide whether to trust a rule. Before you play 5.Nxe5, write down every reason not to play it. Then check each reason against the actual board. Which of your reasons were rules, and which were facts about this position?

Then set up the same position with a black knight on c6 and the pawn back on c7, as in the probable original. Play 5.Nxe5 and find Black's best reply. How much does one knight change the verdict? Write your answer down before you check it with an engine, and then write down what you got wrong.