Jenö Bán, Taktik der Endspiele 1954

White to play and draw

1.g4? doesn't work, because after 1...h3 the white king is shut out of the corner: 2.g5 Kxg5 3.Kf3 Kf5 4.Kf2 Kf4 5.Kf1 Kf3 6.Ke1 Be5 7.Kf1 h2 and Black wins.

The surprising solution is 1.g3+! hxg3+. Who would have thought that White would transform the black h-pawn, which is normally a guarantee for the "wrong bishop" draw, into a dangerous, protected g-pawn. 2.Kg2 Bg1 [naturally 2...Kg4 3.Kh1 Kf4 4.Kg2 is a draw] 3.Kxg1 Kf3 4.Kf1 g2+ 5.Kg1 Kg3 stalemate.

Henri Rinck, 1.hm Boedapest 1911

White to play and draw

1.Kg3! The only move that holds the draw. 1...h5 2.e4! Zugzwang. If it were White to move here Black would easily win. 2...Kg1 3.e5! Forcing the black pawn to capture and move to a square it really doesn't want to be. 3...dxe5 stalemate.

F. Lazard, La Stratégie 1902

White to play and draw

1.Re1! g2+ 2.Kf2+. Precise moves, that leave Black with four options:

In the last variation was in the original solution given by the author, with Black being stalemated, symmetrically with the other underpromotions. But if the rook moves along the e-file White actually win! 3.Re3 Ne2 (or 3...Nf3 4.Kxf3 Kg1 5.Re1#) 4.Re8 Ng1 5.Kg3 Ne2+ (5...Nf3 6.Kxf3 Kg1 7.Re1#) 6.Rxe2 Kg1 7.Re1#.

C. De Feijter, Tijdschrift v.d. KNSB 1941

White to play and draw

1.g7 Re1+ 2.Bd1! Re8. Wasn't this the continuation in which Black could move the rook to g8, capture the g-pawn and win? Yes, but with the following charming twist: 3.Bh5 Rg8 4.Bf7 Rxg7 5.g6!

Take a look at that: the black rook is caught in a cage from which it will never be able to escape. The position is obviously a draw.

A number of readers wrote in to complain that in the final position Fritz will show "Black is winning". Acutally the program displays 4.6 pawns for Black. It is essentially saying that Black is almost a rook up. From the fact that it does not display a 0.00 value (= draw) we must not conclude that there is a terrible flaw or weakness in the program. In fact you should note that in the original position Fritz takes less than ten seconds to the entire drawing line (1.g7 Re1+ 2.Bd1 Re8 3.Bh5 Rg8 4.Bf7 Rxg7 5.g6 etc.) and sticks with it. How many humans, who understand the final position, would find the drawing line in a blitz game?