Another extension of the principles in this chapter. Black’s possible checks (Nc3, Qf1, Qxc2) don’t seem to go anywhere, so he experiments with captures and their consequences. Nxb2 is interesting because White’s only way to recapture would be with his king; any exchange that causes the king to move is interesting. Now what do you see in the resulting position? The king and d2 rook are left a square apart on the same rank, in position to be forked by a pawn on c3. Black has no pawn on the c-file. He does, however, have a pawn on the b-file that is close by; if that pawn could capture something drawn onto c3, it could execute the fork. One way to get an enemy piece onto a square, as we have seen, is to put one of your pieces on the square in a threatening way that requires the enemy to recapture there. Best of all is a check. So Black plays Qc3+, a move that attacks both White’s king and queen; White has to play QxQ (White’s king can’t capture the Black queen because the queen’s square is guarded by a pawn; and if White moves his king, his queen gets taken with QxQ). After White’s QxQ, Black plays b4xQ+—forking king and rook with cover from Black’s own rook on c8. Black in effect has traded a knight for a rook. Again, see how the pawn was able to move over to deliver a fork on a different file after Black drew a White piece onto that file that the pawn could capture.
Now remember that all this assumes White replies to Black’s initial Nxb2 with KxN. White doesn’t have to do that; he can skip the recapture, and indeed is better off doing so—in which case Black wins a pawn rather than the exchange. It is another case where the threat of an eventual fork, if appreciated by both sides, leads to indirect gains.