We have been considering cases where the player facing a relative pin tries to break it by moving the screened piece—in other words, the piece at the rear of the pin. Now we consider another scenario: the possibility of breaking a pin by moving the relatively pinned piece and leaving the piece behind it exposed to capture.
Here Black’s knight on c3 is pinned to his queen by White's bishop, and the knight is attacked three times but defended just once. Does Black have a defense? He needs to find a move with one of the two pieces in the pin—his knight or his queen—that makes a threat; he seeks a threat severe enough to force White to let the other piece in the pin go free on the next turn. The queen has no such threats, but it often is easier to create them with a piece like a knight because knights are more expendable and so can make bold attacks with less fear. So Black examines the knight's circle of moves in search of one that makes a threat great enough to offset the loss of his queen that he risks if the knight moves. It probably would need to be a threat against White’s king or queen. The knight can’t check White’s king because they have one square between them on the same diagonal, a non-working pattern familiar from our work on knight forks. But it can attack the queen with NxN, which also has the virtue of being a capture. If White recaptures, Black moves his queen and the pin has been dissolved with no loss. If White takes Black’s queen, Black takes White’s queen—and has won a piece, since White's queen is unprotected.
Black could play Nb5 here, planning to reply to White's BxQ with his own NxQ. The problem then is that White's queen still is protected (since Black hasn't taken out the knight on e2), so White gets to finish the sequence with NxN and gains a piece.