White’s more advanced knight at f5 has no good forks; the only Black pieces on the knight's square color are the king and bishop, which can’t be reached from any square within range. White’s other knight, though, is on a dark square (d4), and so are Black’s queen and one of his rooks. So how about Nxc6? The square is protected by Black’s bishop, which can’t be pinned, captured, or distracted. But still: if White plays Nxc6 and Black replies with BxN, what then is possible? Imagine White’s d4 knight off the board and Black’s bishop at c6, and then examine every check. There are two: Nxh6+, which doesn't help, and Ne7+—which forks the king and bishop. Then it gets even better: Black has no alternative but to move his king; White then plays NxB, winning the bishop and inflicting yet another fork, this time of Black’s queen and rook. Now it's safe because the bishop is gone.

A lesson of the position: anytime a sequence calls for a knight to move in enemy territory, think about whether its move might be a (or another) fork. Remember that every move by a knight changes the color of the square on which it sits, and thus may radically change its forking prospects.